base-command (bc)¶
Base Command Commands
usage: ngc base-command [--debug] [--format_type <fmt>] [-h] {quickstart} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
base-command¶
- {quickstart}
Possible choices: ace, datamover, dataset, dm, job, qs, quickstart, resource, result, workspace
Sub-commands¶
ace¶
ACE Commands
ngc base-command ace [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
ace¶
Possible choices: info, list, usage
Sub-commands¶
list¶
List each ACE accessible with the current configuration.
ngc base-command ace list [--column <column>] [--debug] [--format_type <fmt>]
[-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is name[=ACE]. Valid columns are id[=Id], description[=Description], instances[=Instances]. Use quotes with spaces. Multiple column arguments are allowed.
info¶
Get ACE details for the given ACE name.
ngc base-command ace info [--debug] [--format_type <fmt>] [-h] <ace name>
Positional Arguments¶
- <ace name>
ACE Name
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
usage¶
[DEPRECATED] Get resource usage information about an ACE.
ngc base-command ace usage [--debug] [--format_type <fmt>]
[--only-unavailable]
[--resource-type {CPU,GPU,MIG}] [-h]
<ace name>
Positional Arguments¶
- <ace name>
ACE Name
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --only-unavailable
Only show items that have unavailable resources.
- --resource-type
Possible choices: CPU, GPU, MIG
Only show items of this resource type.
job¶
Job Commands
ngc base-command job [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
job¶
Possible choices: attach, exec, get-json, info, kill, list, log, preempt, resume, run, telemetry, update
Sub-commands¶
list¶
List all jobs belonging to the user in the last week, filtered by configured ACE and team name. You can also specify the time range using up to two of the options --begin-time, --end-time, and --duration. Acceptable combinations include: --begin-time <t> --end-time <t> (time range is between begin-time and end-time), --begin-time <t> --duration <t> (time range is for a period specified by duration after begin-time), --end-time <t> --duration <t> (time range is for a period specified by duration up to end-time), --end-time <t> (time range is a period of 7 days before end-time), --begin-time <t> (time range is between begin-time and now), --duration <t> (time range is the specified amount of time before now)
ngc base-command job list [--all] [--begin-time <t>] [--column <column>]
[--debug] [--duration <t>] [--end-time <t>]
[--exclude-label <label>] [--format_type <fmt>]
[--interval <num>] [--label <label>] [--long]
[--priority <priority>] [--refresh] [--status <s>]
[-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --all
(For administrators only) Show all jobs across all users.
- --long
Include additional information in the job status table.
- --duration
Specifies the duration of time, either after begin-time or before end-time, for listing jobs created. Format: [nD][nH][nM][nS]. Default: 7 days
- --end-time
Specifies the period end time for listing jobs created. Format: [yyyy-MM-dd::HH:mm:ss]. Default: now
- --begin-time
Specifies the start time for listing jobs created. Format: [yyyy-MM-dd::HH:mm:ss].
- --refresh
Enables refreshing of list.
- --interval
Refresh interval in seconds. Allowed range [5-300] Default: 5
- --status
Possible choices: CANCELED, CREATED, FAILED, FAILED_RUN_LIMIT_EXCEEDED, FINISHED_SUCCESS, IM_INTERNAL_ERROR, INFINITY_POOL_MISSING, KILLED_BY_ADMIN, KILLED_BY_SYSTEM, KILLED_BY_USER, PENDING_STORAGE_CREATION, PENDING_TERMINATION, PREEMPTED, PREEMPTED_BY_ADMIN, QUEUED, REQUESTING_RESOURCE, RESOURCE_CONSUMPTION_REQUEST_IN_PROGRESS, RESOURCE_GRANTED, RESOURCE_GRANT_DENIED, RESOURCE_LIMIT_EXCEEDED, RESOURCE_RELEASED, RUNNING, STARTING, TASK_LOST, UNKNOWN
Filter jobs listed according to input status. Options: ['CANCELED', 'CREATED', 'FAILED', 'FAILED_RUN_LIMIT_EXCEEDED', 'FINISHED_SUCCESS', 'IM_INTERNAL_ERROR', 'INFINITY_POOL_MISSING', 'KILLED_BY_ADMIN', 'KILLED_BY_SYSTEM', 'KILLED_BY_USER', 'PENDING_STORAGE_CREATION', 'PENDING_TERMINATION', 'PREEMPTED', 'PREEMPTED_BY_ADMIN', 'QUEUED', 'REQUESTING_RESOURCE', 'RESOURCE_CONSUMPTION_REQUEST_IN_PROGRESS', 'RESOURCE_GRANTED', 'RESOURCE_GRANT_DENIED', 'RESOURCE_LIMIT_EXCEEDED', 'RESOURCE_RELEASED', 'RUNNING', 'STARTING', 'TASK_LOST', 'UNKNOWN']
- --column
Specify output column as column[=header], header is optional, default is id[=Id]. Valid columns are replicas[=Replicas], name[=Name], submitted[="Submitted By"], status[=Status], details[="Status Details"], type[="Status Type"], created[=Created], started[=Started], ended[=Ended], termination[=Termination], ace[=Ace], team[=Team], org[=Org], instance[="Instance Name"], duration[=Duration], reason[="Termination Reason"], labels[=Labels], locked[="Labels Locked"], order[=Order], priority[=Priority]. Use quotes with spaces. Multiple column arguments are allowed.
- --label
Filter listed jobs by the label. Multiple label arguments are allowed, support standard Unix shell-style wildcards like '*' and '?'.
- --exclude-label
Exclude listed jobs by the label. Multiple exclude label arguments are allowed, support standard Unix shell-style wildcards like '*' and '?'. Filters jobs without labels.
- --priority
Possible choices: HIGH, LOW, NORMAL
Filter jobs listed according to priority. Options: ['HIGH', 'LOW', 'NORMAL']
preempt¶
Preempt the job. This begins a graceful shutdown of a job. Once preempted, the job will remain in a PREEMPTED state until acted on to resume or kill the job.
ngc base-command job preempt [--debug] [--format_type <fmt>] [-h] <job id>
Positional Arguments¶
- <job id>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
resume¶
Resume the preempted job. This takes the job in PREEMPTED state and resubmits the job in the queue.
ngc base-command job resume [--debug] [--format_type <fmt>] [-h] <job id>
Positional Arguments¶
- <job id>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
run¶
Submit a new job. ACE must be set to run this command.
ngc base-command job run [--array-type <type>] [--clone <jobid>]
[--coscheduling] [--datasetid <id>] [--debug]
[--entrypoint <entry>] [--env-var <key:value>]
[--experiment-flow-type <type>]
[--experiment-name <type>]
[--experiment-project-name <type>]
[--format_type <fmt>] [--label <label>]
[--lock-label] [--min-availability <num>]
[--min-timeslice <t>] [--network <type>]
[--order <order>] [--preempt <class>]
[--priority <priority>] [--replicas <num>]
[--result <mntpt>]
[--secret <secret[:key_name:alias_name]>]
[--shell [CMD]] [--start-deadline <t>]
[--topology-constraint <specifier>]
[--total-runtime <t>] [--use-image-entrypoint]
[--waitend] [--waitrun] [-c <c>] [-d <desc>]
[-f <file>] [-h] [-i <url>] [-in <type>] [-n <name>]
[-p <port>] [-w <wkspce>]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -n, --name
Set a job name.
- -i, --image
Set a docker image URL.
- -f, --file
Submit a new job using a JSON File (other arguments will override corresponding json values).
- -c, --commandline
Provide a command for the job to run.
- --entrypoint
Overwrite the default ENTRYPOINT set by the image.
- --use-image-entrypoint
Use the ENTRYPOINT defined in the image manifest
- -d, --description
Set a job description.
- --datasetid
Specify a dataset ID with container mountpoint or no-mount for a object based dataset(format: dataset-id:mountPoint or dataset-id:no-mount) to be bound to the job. This can be supplied multiple times. If no-mount is provided, the job will try to the fetch the dataset through storage specific protocol(e.g. s3 protocol).
- -in, --instance
Instance type.
- --replicas
Specifies the number of child replicas created for multi-node parallel job.
- --array-type
Possible choices: HOROVOD, MPI, PARALLEL, PYTORCH
Specifies the type of Job. Choices: MPI, PARALLEL, PYTORCH, HOROVOD.
- --coscheduling
Specifies if coscheduling would be allowed for the multi-node parallel job submission.
- --min-availability
Minimum replicas that need to be scheduled to start a multi-node job. --min-availability is not allowed with --coscheduling.
- --network
Possible choices: ETHERNET, INFINIBAND
Specify the information pertaining to network or switch. Choices: INFINIBAND, ETHERNET. Default: ETHERNET
- --topology-constraint
Possible choices: any, pack
Specifies a topology constraint for the job. Only available for multi-node jobs. Choices: pack, any.
- -p, --port
Set ports to open on the docker container. Ports on the host do not need to be specified. Allowed range for containerPort is [1-1728][1730-65535]. Multiple port arguments are allowed. ACEs that allow exposed port supports the format name:containerPort/protocol. Choices for the protocol are: TCP, UDP, HTTPS, GRPC. Name must contain only alphanumeric characters, start with an alphabet and be no more than 10 chars. HTTPS and GRPC protocols do not support name. HTTPS is applied if the protocol is not specfied.
- --result
Mount point for the job result.
- --preempt
Possible choices: RESTARTABLE, RESUMABLE, RUNONCE
Specify the job class for preemption and scheduling behavior. One of RESUMABLE, RESTARTABLE, or RUNONCE (default for non-shell jobs).
- --total-runtime
Maximum cumulative duration (in the format [nD][nH][nM][nS]) the job is in the RUNNING state before it gets gracefully shut down by the system.
- --min-timeslice
Minimum duration (in the format [nD][nH][nM][nS]) the job is expected (not guaranteed) to be in the RUNNING state once scheduled to assure forward progress.
- --waitrun
The CLI will block until the job reaches a RUNNING status or user exits with Ctrl-C.
- --waitend
The CLI will block until the job reaches a terminal status or user exits with Ctrl-C.
- --start-deadline
Maximum duration (in the format [nD][nH][nM][nS]) the job will have to reach a RUNNING status before it is automatically killed. May only be used with --shell. Default: 6m
- --shell
Automatically exec into the running container once the job starts with an optionally supplied command (defaults to /bin/bash). If --commandline is not supplied with this option, 'sleep' will be used to hold the container open. --total-runtime controls the duration of the sleep command and defaults to 8H.
- -w, --workspace
Specify the workspace to be bound to the job. (format: <workspace-id|workspace-name>:<mountpoint>:<mount-mode>). <mount-mode> can take values RW (read-write), RO (read-only) (default: RW). Multiple workspace arguments are allowed.
- --clone
Submit a new job by cloning an existing job (other arguments will override corresponding values).
- --label
Specify labels for the job. Multiple label arguments are allowed. Labels must start with alphabetic characters or '_' and valid characters are alphanumeric and '_'. It must be no more than 256 characters. Reserved labels start with '_'. System labels start with '__' and only admins can assign or remove system labels. A maximum of 20 user, reserved or system labels are allowed.
- --lock-label
Lock labels for the job. Default is unlocked.
- --order
Specify order for the job. Default is 50. Job order is from 1 to 99.
- --priority
Possible choices: HIGH, LOW, NORMAL
Specify priority for the job. Default is NORMAL. Choices ['HIGH', 'LOW', 'NORMAL']
- --secret
Specify secret name for the job. Multiple secret arguments are allowed. Unless specified all key value pairs will be included in the job. Optionally per key-value pair, overrides of the key are available.
- --env-var
A custom env variable to add to job in the form of a key-pairA key name must be between 1-63 characters and contain letters, numbers or ./-_ May be used multiple times in the same command.
- --experiment-flow-type
Possible choices: NONE, WANDB
Third-party experiment flow type which will be used to export specific environment variables to associate the project/experiment name to the job and decide the format of the experimentation tracking URLSpecifies the type of Job. Choices: NONE, WANDB.
- --experiment-project-name
Third-party project/environment name to associate the current job/run
- --experiment-name
Optional Third-party experiment name to group the jobs/runs
info¶
Get job details by job ID.
ngc base-command job info [--debug] [--format_type <fmt>] [-h]
<job_id>[:replica_id]>
Positional Arguments¶
- <job_id>[:replica_id]>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
attach¶
Attach to the container running the provided job.
ngc base-command job attach [--debug] [--format_type <fmt>] [-h]
<job_id>[:replica_id]>
Positional Arguments¶
- <job_id>[:replica_id]>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
exec¶
Exec to the container running the provided job.
ngc base-command job exec [--commandline] [--debug] [--detach]
[--format_type <fmt>] [-h]
<job_id>[:replica_id]>
Positional Arguments¶
- <job_id>[:replica_id]>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --commandline
The command you want to execute with exec.
- --detach
Detach from the exec command.
kill¶
Submit a request to kill jobs by job ID.
ngc base-command job kill [--debug] [--dry-run] [--format_type <fmt>]
[--reason <reason>] [--status <s>] [-h]
<jobid|jobrange|joblist>
Positional Arguments¶
- <jobid|jobrange|joblist>
Job ID(s). Valid Examples: '1-5', '333', '1, 2', '1,10-15'
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --status
Possible choices: CREATED, PENDING_STORAGE_CREATION, PENDING_TERMINATION, PREEMPTED, PREEMPTED_BY_ADMIN, QUEUED, REQUESTING_RESOURCE, RESOURCE_CONSUMPTION_REQUEST_IN_PROGRESS, RESOURCE_GRANTED, RUNNING, STARTING
Kill jobs that match the provided status. Multiple --status flags will OR together. Options: ['CREATED', 'PENDING_STORAGE_CREATION', 'PENDING_TERMINATION', 'PREEMPTED', 'PREEMPTED_BY_ADMIN', 'QUEUED', 'REQUESTING_RESOURCE', 'RESOURCE_CONSUMPTION_REQUEST_IN_PROGRESS', 'RESOURCE_GRANTED', 'RUNNING', 'STARTING']
- --dry-run
List jobs to be killed without performing the action.
- --reason
Reason to terminate the job (required for administrators).
telemetry¶
List telemetry data for the given job.
ngc base-command job telemetry [--debug] [--format_type <fmt>]
[--interval-time <t>] [--interval-unit <u>]
[--statistics <form>] [--type <type>] [-h]
<<job id>[:replica_id]>
Positional Arguments¶
- <<job id>[:replica_id]>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --interval-unit
Possible choices: HOUR, MINUTE, SECOND
Data collection interval unit. Options: ['HOUR', 'MINUTE', 'SECOND']. Default: MINUTE
- --interval-time
Data collection interval time value. Default: 1
- --type
Possible choices: APPLICATION_TELEMETRY, CPU_UTILIZATION, GPU_FB_USED, GPU_FI_PROF_DRAM_ACTIVE, GPU_FI_PROF_PCIE_RX_BYTES, GPU_FI_PROF_PCIE_TX_BYTES, GPU_FI_PROF_PIPE_TENSOR_ACTIVE, GPU_NVLINK_BANDWIDTH_TOTAL, GPU_NVLINK_RX_BYTES, GPU_NVLINK_TX_BYTES, GPU_POWER_USAGE, GPU_UTILIZATION, MEM_UTILIZATION, NETWORK_LOCAL_STORAGE_RAID_SIZE_BYTES, NETWORK_LOCAL_STORAGE_RAID_TOTAL_READ_BYTES, NETWORK_LOCAL_STORAGE_RAID_TOTAL_WRITE_BYTES, NETWORK_LOCAL_STORAGE_RAID_USAGE_BYTES, NETWORK_LOCAL_STORAGE_ROOT_SIZE_BYTES, NETWORK_LOCAL_STORAGE_ROOT_TOTAL_READ_BYTES, NETWORK_LOCAL_STORAGE_ROOT_TOTAL_WRITE_BYTES, NETWORK_LOCAL_STORAGE_ROOT_USAGE_BYTES, NETWORK_RDMA_PORT_RATE_RX_BYTES, NETWORK_RDMA_PORT_RATE_TX_BYTES, NETWORK_RDMA_PORT_RX_BYTES, NETWORK_RDMA_PORT_TX_BYTES, NETWORK_RX_BYTES_TOTAL, NETWORK_STORAGE_TOTAL_READ_BYTES, NETWORK_STORAGE_TOTAL_WRITE_BYTES, NETWORK_STORAGE_TRANSPORT_RX_BYTES, NETWORK_STORAGE_TRANSPORT_TX_BYTES, NETWORK_TX_BYTES_TOTAL
A telemetry type to report. Options: ['APPLICATION_TELEMETRY', 'CPU_UTILIZATION', 'GPU_FB_USED', 'GPU_FI_PROF_DRAM_ACTIVE', 'GPU_FI_PROF_PCIE_RX_BYTES', 'GPU_FI_PROF_PCIE_TX_BYTES', 'GPU_FI_PROF_PIPE_TENSOR_ACTIVE', 'GPU_NVLINK_BANDWIDTH_TOTAL', 'GPU_NVLINK_RX_BYTES', 'GPU_NVLINK_TX_BYTES', 'GPU_POWER_USAGE', 'GPU_UTILIZATION', 'MEM_UTILIZATION', 'NETWORK_LOCAL_STORAGE_RAID_SIZE_BYTES', 'NETWORK_LOCAL_STORAGE_RAID_TOTAL_READ_BYTES', 'NETWORK_LOCAL_STORAGE_RAID_TOTAL_WRITE_BYTES', 'NETWORK_LOCAL_STORAGE_RAID_USAGE_BYTES', 'NETWORK_LOCAL_STORAGE_ROOT_SIZE_BYTES', 'NETWORK_LOCAL_STORAGE_ROOT_TOTAL_READ_BYTES', 'NETWORK_LOCAL_STORAGE_ROOT_TOTAL_WRITE_BYTES', 'NETWORK_LOCAL_STORAGE_ROOT_USAGE_BYTES', 'NETWORK_RDMA_PORT_RATE_RX_BYTES', 'NETWORK_RDMA_PORT_RATE_TX_BYTES', 'NETWORK_RDMA_PORT_RX_BYTES', 'NETWORK_RDMA_PORT_TX_BYTES', 'NETWORK_RX_BYTES_TOTAL', 'NETWORK_STORAGE_TOTAL_READ_BYTES', 'NETWORK_STORAGE_TOTAL_WRITE_BYTES', 'NETWORK_STORAGE_TRANSPORT_RX_BYTES', 'NETWORK_STORAGE_TRANSPORT_TX_BYTES', 'NETWORK_TX_BYTES_TOTAL']. Default: None
- --statistics
Possible choices: MAX, MEAN, MIN
Statistical form of the data to report. Options: ['MAX', 'MEAN', 'MIN']. Default: MEAN
get-json¶
Generate a json file containing the given job. This can be used to submit that job.
ngc base-command job get-json [--debug] [--format_type <fmt>] [-h] <job id>
Positional Arguments¶
- <job id>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
update¶
Update a job's labels.
ngc base-command job update [--clear-label] [--debug] [--format_type <fmt>]
[--label <label>] [--lock-label]
[--remove-label REMOVE_LABEL] [--unlock-label]
[-h]
<job id>
Positional Arguments¶
- <job id>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --label
Specify labels for the job. Multiple label arguments are allowed. Labels must start with alphabetic characters or '_' and valid characters are alphanumeric and '_'. It must be no more than 256 characters. Reserved labels start with '_'. System labels start with '__' and only admins can assign or remove system labels. A maximum of 20 user, reserved or system labels are allowed.
- --remove-label
Remove a label. Multiple remove label arguments are allowed, support standard Unix shell-style wildcards like '*' and '?'.
- --clear-label
Remove all labels for the job.
- --lock-label
Lock Labels.
- --unlock-label
Unlock Labels.
log¶
Print a job's log.
ngc base-command job log [--debug] [--format_type <fmt>] [--head]
[--lines <lines>] [--tail] [-h]
<job_id>[:replica_id]>
Positional Arguments¶
- <job_id>[:replica_id]>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --head
Print the first part of the log file. Default is 10 lines.
- --lines
Specify the number of lines to print. Must specify --head or --tail.
- --tail
Print the last part of the log file. Default is 10 lines.
dataset¶
Data Commands
ngc base-command dataset [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
dataset¶
Possible choices: convert, download, export, import, info, list, remove, revoke-share, share, upload
Sub-commands¶
export¶
Dataset Export Commands
ngc base-command dataset export [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
export¶
Possible choices: info, list, run
Sub-commands¶
list¶
List all dataset export jobs.
ngc base-command dataset export list [--column <column>] [--debug]
[--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is id[=Id]. Valid columns are id[=Id], source[=Source], destination[=Destination], status[=Status], start_time[="Start time"], end_time[="End time"]. Use quotes with spaces. Multiple column arguments are allowed.
run¶
Export a dataset from the ACE into an object store.
ngc base-command dataset export run [--account-name <account-name>]
[--bucket <bucket>]
[--container <container>] [--debug]
[--endpoint <endpoint>]
[--format_type <fmt>]
[--instance <instance>]
[--prefix <prefix>] --protocol <protocol>
[--region <region>] --secret <secret>
[--service-url <service-url>] [-h]
<dataset>
Positional Arguments¶
- <dataset>
Dataset ID to be exported.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --endpoint
S3 endpoint. Only applies when --protocol is s3.
- --bucket
S3 bucket name. Only applies when --protocol is s3.
- --region
S3 region (optional). Default: us-east-1. Only applies when --protocol is s3.
- --account-name
Azure Blob account name. Only applies when --protocol is azureblob.
- --container
Azure Blob container name. Only applies when --protocol is azureblob.
- --service-url
Azure Blob service url (optional). Only applies when --protocol is azureblob.
- --prefix
Object prefix. Enables copying a subset of all objects in a location
- --instance
Instance to use for the data export.
Required named arguments¶
- --protocol
Possible choices: azureblob, s3, url
Access protocol for the destination. Options: s3, url, azureblob.
- --secret
NGC Secret object to use.
info¶
Status of the dataset export job.
ngc base-command dataset export info [--debug] [--format_type <fmt>] [-h]
<job_id>
Positional Arguments¶
- <job_id>
Job ID.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
import¶
Dataset Import Commands
ngc base-command dataset import [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
import¶
Possible choices: finish, info, list, start
Sub-commands¶
list¶
List all dataset import jobs
ngc base-command dataset import list [--column <column>] [--debug]
[--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is id[=Id]. Valid columns are id[=Id], source[=Source], destination[=Destination], status[=Status], start_time[="Start time"], end_time[="End time"]. Use quotes with spaces. Multiple column arguments are allowed.
start¶
Start import of a dataset from an object store into the ACE.
ngc base-command dataset import start [--account-name <account-name>]
[--bucket <bucket>]
[--container <container>] [--debug]
[--endpoint <endpoint>]
[--format_type <fmt>] --instance
<instance> [--prefix <prefix>]
--protocol <protocol>
[--region <region>] --secret <secret>
[--service-url <service-url>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --endpoint
S3 endpoint. Only applies when --protocol is s3.
- --bucket
S3 bucket name. Only applies when --protocol is s3.
- --prefix
Object prefix. Enables copying a subset of all objects in a location
- --region
S3 region (optional). Default: us-east-1. Only applies when --protocol is s3.
- --account-name
Azure Blob account name. Only applies when --protocol is azureblob.
- --container
Azure Blob container name. Only applies when --protocol is azureblob.
- --service-url
Azure Blob service url (optional). Only applies when --protocol is azureblob.
Required named arguments¶
- --protocol
Possible choices: azureblob, s3, url
Access protocol for the source. Options: s3, url, azureblob.
- --secret
NGC Secret object to use.
- --instance
Instance to use for the data import.
info¶
Status of the import job.
ngc base-command dataset import info [--debug] [--format_type <fmt>] [-h]
<job_id>
Positional Arguments¶
- <job_id>
Job ID.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
finish¶
Finish a dataset import.
ngc base-command dataset import finish [--debug] [--desc <desc>]
[--format_type <fmt>]
[--from-dataset <id>] [--name <name>]
[-h]
<job_id>
Positional Arguments¶
- <job_id>
Job ID.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --from-dataset
Dataset to import metadata from. Used during ACE to ACE copies, typically after having run a dataset export. Imported metadata: name, description, sharing information.
- --name
Dataset Name. If used together with --from-dataset, --name takes precedence.
- --desc
Dataset Description. If used together with --from-dataset, --desc takes precedence.
list¶
List all accessible datasets. Set ACE and team will be used to filter list.
ngc base-command dataset list [--all] [--column <column>] [--debug]
[--format_type <fmt>] [--name <name>] [--owned]
[--prepopulated] [--status <status>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --owned
Include only owned datasets.
- --prepopulated
Include only pre-populated datasets.
- --name
Include only datasets with name <name>, wildcards '*' and '?' are allowed.
- --status
Possible choices: COMPLETED, CONVERTED, CREATED, DELETED, INITIALIZING
Include only datasets with status <status>. Options: ['COMPLETED', 'CONVERTED', 'CREATED', 'DELETED', 'INITIALIZING'].
- --all
(For administrators only) Show all datasets across all users.
- --column
Specify output column as column[=header], header is optional, default is uid[=Id]. Valid columns are id[="Integer Id"], name[=Name], org[=Org], team[=Team], modified[="Modified Date"], created[="Created Date"], creator[="Creator UserName"], description[=Description], shared[=Shared], owned[=Owned], ace[=Ace], status[=Status], size[=Size], prepop[=Pre-pop]. Use quotes with spaces. Multiple column arguments are allowed.
info¶
Retrieve details of a dataset given dataset ID
ngc base-command dataset info [--debug] [--files] [--format_type <fmt>] [-h]
<dataset id>
Positional Arguments¶
- <dataset id>
Dataset ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --files
List files in addition to details for a dataset. Default value is False
upload¶
Upload a dataset to a given ACE. Dataset will be local to the set ACE.
ngc base-command dataset upload [--debug] [--desc <desc>] [--dry-run]
[--format_type <fmt>] [--omit-links]
[--share [<team>]] [--source <path>]
[--threads <t>] [-h] [-y]
<dataset>
Positional Arguments¶
- <dataset>
Dataset Name or ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --desc
Dataset Description
- --source
Path to the file(s) to be uploaded. Default: .
- --threads
Number of threads to be used while uploading the dataset (default: 12, max: 21)
- -y, --yes
Automatically say yes to all interactive questions.
- --omit-links
Do not follow symbolic links.
- --dry-run
List file paths, total upload size and file count without performing the upload.
- --share
Share the dataset with a team after upload. Can be used multiple times. If no team is specified, the currently set team will be used.
download¶
Download datasets by ID.
ngc base-command dataset download [--debug] [--dest <path>] [--dir <wildcard>]
[--dry-run] [--exclude <wildcard>]
[--file <wildcard>] [--format_type <fmt>]
[--resume <resume>] [--zip] [-h]
<dataset id>
Positional Arguments¶
- <dataset id>
Dataset ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Specify the path to store the downloaded files. Default: .
- --file
Specify individual files to download from the dataset. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..) May be used multiple times in the same command.
- --resume
Resume the download for the dataset. Specify the file name saved by the download. Files will be downloaded to the directory of the file name.
- --dir
Specify directories to download from dataset. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..) May be used multiple times in the same command.
- --zip
Download the entire dataset directory as a zip file.
- --exclude
Exclude files or directories from the downloaded dataset. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..). May be used multiple times in the same command.
- --dry-run
List total size of the download without performing the download.
remove¶
Remove a dataset.
ngc base-command dataset remove [--debug] [--format_type <fmt>] [-h] [-y]
<datasetid|datasetrange|datasetlist>
Positional Arguments¶
- <datasetid|datasetrange|datasetlist>
Dataset ID(s). Valid Examples: '1-5', '333', '1,2', '1,10-15'. Do not include any spaces between IDs. Dataset range is not supported while using Data Platform API.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
convert¶
Convert data from a variety of sources to a dataset in the set ACE.
ngc base-command dataset convert [--debug] [--desc <desc>]
[--format_type <fmt>] --from-result <id> [-h]
<name>
Positional Arguments¶
- <name>
Dataset Name
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --desc
Provide a description for the dataset.
Required named arguments¶
- --from-result
Job result to convert to a dataset. Must be in the same ACE as target. Result files are no longer available after conversion.
workspace¶
Workspace Commands
ngc base-command workspace [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
workspace¶
Possible choices: create, download, export, import, info, list, mount, remove, revoke-share, set, share, unmount, upload
Sub-commands¶
export¶
Workspace Export Commands
ngc base-command workspace export [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
export¶
Possible choices: info, list, run
Sub-commands¶
list¶
List all workspace export jobs.
ngc base-command workspace export list [--column <column>] [--debug]
[--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is id[=Id]. Valid columns are id[=Id], source[=Source], destination[=Destination], status[=Status], start_time[="Start time"], end_time[="End time"]. Use quotes with spaces. Multiple column arguments are allowed.
run¶
Export a workspace from the ACE into an object store.
ngc base-command workspace export run [--account-name <account-name>]
[--bucket <bucket>]
[--container <container>] [--debug]
[--endpoint <endpoint>]
[--format_type <fmt>]
[--instance <instance>]
[--prefix <prefix>] --protocol
<protocol> [--region <region>] --secret
<secret> [--service-url <service-url>]
[-h]
<workspace>
Positional Arguments¶
- <workspace>
Workspace ID to be exported.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --endpoint
S3 endpoint. Only applies when --protocol is s3.
- --bucket
S3 bucket name. Only applies when --protocol is s3.
- --prefix
Object prefix. Enables copying a subset of all objects in a location
- --region
S3 region (optional). Default: us-east-1. Only applies when --protocol is s3.
- --account-name
Azure Blob account name. Only applies when --protocol is azureblob.
- --container
Azure Blob container name. Only applies when --protocol is azureblob.
- --service-url
Azure Blob service url (optional). Only applies when --protocol is azureblob.
- --instance
Instance to use for the data export.
Required named arguments¶
- --protocol
Possible choices: azureblob, s3, url
Access protocol for the destination. Options: s3, url, azureblob.
- --secret
NGC Secret object to use.
info¶
Status of the workspace export job.
ngc base-command workspace export info [--debug] [--format_type <fmt>] [-h]
<job_id>
Positional Arguments¶
- <job_id>
Job ID.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
import¶
Workspace Import Commands
ngc base-command workspace import [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
import¶
Possible choices: info, list, run
Sub-commands¶
list¶
List all workspace import jobs
ngc base-command workspace import list [--column <column>] [--debug]
[--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is id[=Id]. Valid columns are id[=Id], source[=Source], destination[=Destination], status[=Status], start_time[="Start time"], end_time[="End time"]. Use quotes with spaces. Multiple column arguments are allowed.
run¶
Import of a workspace from an object store into the ACE.
ngc base-command workspace import run [--account-name <account-name>]
[--bucket <bucket>]
[--container <container>] [--debug]
[--desc <desc>] [--endpoint <endpoint>]
[--format_type <fmt>]
[--from-workspace <id>] --instance
<instance> [--name <name>]
[--prefix <prefix>] --protocol
<protocol> [--region <region>] --secret
<secret> [--service-url <service-url>]
[-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --endpoint
S3 endpoint. Only applies when --protocol is s3.
- --bucket
S3 bucket name. Only applies when --protocol is s3.
- --prefix
Object prefix. Enables copying a subset of all objects in a location
- --region
S3 region (optional). Default: us-east-1. Only applies when --protocol is s3.
- --account-name
Azure Blob account name. Only applies when --protocol is azureblob.
- --container
Azure Blob container name. Only applies when --protocol is azureblob.
- --service-url
Azure Blob service url (optional). Only applies when --protocol is azureblob.
- --from-workspace
Workspace to import metadata from. Used during ACE to ACE copies, typically after having run a workspace export. Imported metadata: name, description, sharing information.
- --name
Workspace Name. If used together with --from-workspace, --name takes precedence.
- --desc
Workspace Description. If used together with --from-workspace, --desc takes precedence.
Required named arguments¶
- --protocol
Possible choices: azureblob, s3, url
Access protocol for the source. Options: s3, url, azureblob.
- --secret
NGC Secret object to use.
- --instance
Instance to use for the data import.
info¶
Status of the import job.
ngc base-command workspace import info [--debug] [--format_type <fmt>] [-h]
<job_id>
Positional Arguments¶
- <job_id>
Job ID.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
create¶
Create a new workspace.
ngc base-command workspace create [--debug] [--format_type <fmt>]
[--name <name>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --name
Set the workspace name. This may only be done once.
remove¶
Remove a workspace.
ngc base-command workspace remove [--debug] [--format_type <fmt>] [-h] [-y]
<workspace>
Positional Arguments¶
- <workspace>
Workspace Name or ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
mount¶
Mount an existing workspace.
ngc base-command workspace mount [--control-path] [--debug] [--force]
[--format_type <fmt>] --mode <mode>
[--remote-path <path>] [-h] [-y]
<workspace> <local path>
Positional Arguments¶
- <workspace>
Workspace Name or ID
- <local path>
Local Mount Point Name
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --remote-path
Path on the remote server inside the workspace.
- --force
Force mount; this will remount if there is a broken mount point.
- -y, --yes
Automatically say yes to all interactive questions.
- --control-path
Set contol path to none and enable control master.
Required named arguments¶
- --mode
Possible choices: RO, RW
Mount mode. Valid values: RW, RO
unmount¶
Unmount a workspace.
ngc base-command workspace unmount [--debug] [--format_type <fmt>] [-h]
<local path>
Positional Arguments¶
- <local path>
Local Mount Point Name
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
info¶
Get workspace details.
ngc base-command workspace info [--debug] [--format_type <fmt>] [--show-sftp]
[-h]
<workspace>
Positional Arguments¶
- <workspace>
Workspace Name or ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --show-sftp
Show hostname, port and token for sftp.
set¶
Set name and/or description for a workspace.
ngc base-command workspace set [--debug] [--desc <desc>] [--format_type <fmt>]
[-h] [-n <name>] [-y]
<workspaceid>
Positional Arguments¶
- <workspaceid>
Workspace ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -n, --name
Set a workspace name. This may be done once.
- --desc
Set a workspace description. This may be done once.
- -y, --yes
Automatically say yes to all interactive questions.
list¶
List all accessible workspaces. Current ACE and team will be used to filter the output.
ngc base-command workspace list [--all] [--column <column>] [--debug]
[--format_type <fmt>] [--name <name>]
[--owned] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --owned
Include only owned workspaces.
- --name
Include only workspaces with name <name>; the wildcards '*' and '?' are allowed.
- --all
(For administrators only) Show all workspaces across all users.
- --column
Specify output column as column[=header], header is optional, default is id[=Id]. Valid columns are name[=Name], org[=Org], team[=Team], updated[="Updated Date"], created[="Created Date"], creator[="Creator UserName"], description[=Description], shared[=Shared], owned[=Owned], ace[=Ace], size[=Size]. Use quotes with spaces. Multiple column arguments are allowed.
upload¶
Upload files to a workspace.
ngc base-command workspace upload [--debug] [--destination <path>] [--dry-run]
[--exclude <wildcard>] [--format_type <fmt>]
[--source <path>] [--threads <t>] [-h]
<workspace>
Positional Arguments¶
- <workspace>
Workspace Name or ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --threads
Number of threads to be used while uploading the workspace (default: 12, max: 21)
- --source
Provide the path to the file(s) to be uploaded. Default: .
- --destination
Provide a target directory within the workspace for the upload. Default: /
- --exclude
Exclude files or directories from the source path. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..). May be used multiple times in the same command.
- --dry-run
List file paths, total upload size and file count without performing the upload.
download¶
Download a workspace.
ngc base-command workspace download [--debug] [--dest <path>] [--dir <path>]
[--dry-run] [--file <path>]
[--format_type <fmt>] [--zip] [-h]
<workspace>
Positional Arguments¶
- <workspace>
Workspace Name or ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Specify the path to store the downloaded workspace. Default: .
- --file
Specify file(s) to download. This flag can be used multiple times. If omitted, the entire workspace directory will be downloaded.
- --dir
Specify one or more directories to download. If omitted, the entire workspace directory will be downloaded.
- --zip
Download the entire dataset directory as a zip file.
- --dry-run
List total size of the download without performing the download.
datamover (dm)¶
Data Mover commands to assist copying resources to/from an object storage or to another ACE
ngc base-command datamover [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
datamover¶
Possible choices: enqueue, list, update
Sub-commands¶
enqueue¶
Add copy job(s) to Data Mover queue.
ngc base-command datamover enqueue [--account-name <account-name>]
[--all-from-org] [--all-from-team]
[--bucket <bucket>]
[--container <container>] [--debug]
[--destination-ace <destination-ace>]
[--destination-instance <destination-instance>]
[--endpoint <endpoint>]
[--format_type <fmt>] [--id <id>]
[--manifest <manifest>]
[--origin-ace <origin-ace>]
[--origin-instance <origin-instance>]
[--prefix <prefix>] --protocol <protocol>
[--region <region>] --resource-type
<resource_type> --secret <secret>
[--service-url <service-url>] [-h] [-y]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --endpoint
S3 endpoint. Only applies when --protocol is s3.
- --bucket
S3 bucket name. Only applies when --protocol is s3.
- --region
S3 region (optional). Default: us-east-1. Only applies when --protocol is s3.
- --account-name
Azure Blob account name. Only applies when --protocol is azureblob.
- --container
Azure Blob container name. Only applies when --protocol is azureblob.
- --service-url
Azure Blob service url (optional). Only applies when --protocol is azureblob.
- --prefix
Object prefix. Enables copying a subset of all objects in a location
- --origin-instance
Instance to use for the data export.
- --destination-instance
Instance to use for the data import.
- --origin-ace
Origin ACE name
- --destination-ace
Destination ACE name
- --manifest
Copy all resources from provided manifest file
- --id
Id of the single resource to be copied
- --all-from-org
Copy all resources from organization
- --all-from-team
Copy all resources from team
- -y, --yes
Automatically say yes to all interactive questions.
Required named arguments¶
- --protocol
Possible choices: azureblob, s3, url
Access protocol for the destination. Options: s3, url, azureblob.
- --secret
NGC Secret object to use.
- --resource-type
Possible choices: dataset, workspace
Type of resource to be copied. Options: dataset, workspace
update¶
Move available jobs to the next data movement stage.
ngc base-command datamover update [--debug] [--format_type <fmt>]
[--interval <seconds>] [--loop] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --loop
Run the update command in an endless loop.
- --interval
Interval in seconds for endless loop. Default: 10
list¶
List status of Data Mover jobs.
ngc base-command datamover list [--debug] [--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
quickstart (qs)¶
QuickStart Commands
ngc base-command quickstart [--debug] [--format_type <fmt>] [-h]
{cluster,project} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
quickstart¶
- {cluster,project}
Possible choices: cluster, project
Sub-commands¶
cluster¶
QuickStart Cluster Commands
ngc base-command quickstart cluster [--debug] [--format_type <fmt>] [-h]
{create,info,list,list-instance-types,remove,start,status,stop,update}
...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
cluster¶
- {create,info,list,list-instance-types,remove,start,status,stop,update}
Possible choices: create, info, list, list-instance-types, remove, start, status, stop, update
Sub-commands¶
list-instance-types¶
Show a list of all available instance types
ngc base-command quickstart cluster list-instance-types --cluster-type
{dask,jupyterlab}
[--debug]
[--format_type <fmt>]
[--multinode] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --multinode
Show multinode instance types. Default=False
Required named arguments¶
- --cluster-type
Possible choices: dask, jupyterlab
The type of cluster: choose from ['dask', 'jupyterlab'].
list¶
List clusters.
ngc base-command quickstart cluster list --cluster-type {dask,jupyterlab}
[--column <column>] [--debug]
[--format_type <fmt>] [--org-only]
[--owned] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is name[=Name]. Valid columns are additionalInfo[="Additional Info"], id[=ID], name[=Name], org[=Org], status[=Status], team[=Team], type[=Type]. Use quotes with spaces. Multiple column arguments are allowed.
- --org-only
Don't return clusters created at the team level
- --owned
Only return clusters I own (admin only)
Required named arguments¶
- --cluster-type
Possible choices: dask, jupyterlab
The type of cluster: choose from ['dask', 'jupyterlab'].
info¶
Show information about a cluster
ngc base-command quickstart cluster info [--debug] [--format_type <fmt>] [-h]
cluster_id
Positional Arguments¶
- cluster_id
The ID of the cluster
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
status¶
Show the status of a cluster
ngc base-command quickstart cluster status [--debug] [--format_type <fmt>]
[-h]
cluster_id
Positional Arguments¶
- cluster_id
The ID of the cluster
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
remove¶
Shutdown and delete a cluster
ngc base-command quickstart cluster remove [--debug] [--format_type <fmt>]
[-h]
cluster_id
Positional Arguments¶
- cluster_id
The ID of the cluster
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
stop¶
Stop a cluster
ngc base-command quickstart cluster stop [--debug] [--format_type <fmt>] [-h]
cluster_id
Positional Arguments¶
- cluster_id
The ID of the cluster
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
start¶
Start a cluster
ngc base-command quickstart cluster start [--debug] [--format_type <fmt>] [-h]
cluster_id
Positional Arguments¶
- cluster_id
The ID of the cluster
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
create¶
Create a new cluster
ngc base-command quickstart cluster create
[--additional-open-ports ADDITIONAL_OPEN_PORTS]
[--additional-port-mappings ADDITIONAL_PORT_MAPPINGS]
--cluster-lifetime CLUSTER_LIFETIME
--cluster-type {dask,jupyterlab}
[--conda-packages CONDA_PACKAGES]
--container-image CONTAINER_IMAGE
--data-output-mount-point
DATA_OUTPUT_MOUNT_POINT
[--dataset-mount DATASET_MOUNT]
[--debug]
[--expiry-duration EXPIRY_DURATION]
[--format_type <fmt>]
[--job-order JOB_ORDER]
[--job-priority JOB_PRIORITY]
[--label LABEL] [--labels-locked]
[--min-availability MIN_AVAILABILITY]
[--min-time-slice MIN_TIME_SLICE]
[--multi-node] [--name NAME]
--nworkers NWORKERS
[--options <key:value>]
[--pip-packages PIP_PACKAGES]
[--preempt-class PREEMPT_CLASS]
[--scheduler-dashboard-address SCHEDULER_DASHBOARD_ADDRESS]
[--scheduler-env-var SCHEDULER_ENV_VAR]
--scheduler-instance-type
SCHEDULER_INSTANCE_TYPE
[--scheduler-port SCHEDULER_PORT]
[--scheduler-reserved-gpus SCHEDULER_RESERVED_GPUS]
[--scheduler-startup-script SCHEDULER_STARTUP_SCRIPT]
[--system-packages SYSTEM_PACKAGES]
[--topology-constraint {any,pack}]
[--user-secret <secret[:key_name:alias_name]>]
[--worker-dashboard-address WORKER_DASHBOARD_ADDRESS]
[--worker-env-var WORKER_ENV_VAR]
[--worker-instance-type WORKER_INSTANCE_TYPE]
[--worker-reserved-gpus WORKER_RESERVED_GPUS]
[--worker-startup-script WORKER_STARTUP_SCRIPT]
[--workspace-mount WORKSPACE_MOUNT]
[-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --name
The name of the cluster
- --scheduler-dashboard-address
The dashboard address for the scheduler. Only used for 'dask' cluster type.
- --scheduler-port
The port to use for the scheduler. Only used for 'dask' cluster type.
- --scheduler-startup-script
The startup script for the scheduler
- --scheduler-reserved-gpus
The number of GPUs reserved for the scheduler
- --scheduler-env-var
An environment variable to be set in the scheduler node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--scheduler-env-var' arguments.
- --worker-dashboard-address
The dashboard address for the worker. Only used for 'dask' cluster type.
- --worker-instance-type
The instance type of the worker. Required for 'dask' cluster type; ignored otherwise.
- --worker-startup-script
The startup script for the worker. Only used for 'dask' cluster type.
- --worker-reserved-gpus
The number of GPUs reserved for the worker. Only used for 'dask' cluster type.
- --worker-env-var
An environment variable to be set in the worker node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--worker-env-var' arguments. Only used for 'dask' cluster type.
- --system-packages
List of packages to install on the scheduler and worker using 'apt install' or 'yum install' command. (apt or yum command is chosen depending on the flavour of the Linux image in the Container). You may define more than one package by specifying multiple '--system-packages' arguments.
- --conda-packages
List of packages to install on the scheduler and worker using 'conda install' command. You may define more than one package by specifying multiple '--conda-packages' arguments.
- --pip-packages
List of packages to install on the scheduler and worker using 'pip install' command. You may define more than one package by specifying multiple '--pip-packages' arguments.
- --expiry-duration
The expiry duration for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
- --additional-open-ports
(deprecated; use additional-port-mappings instead) Any additional ports to open for the cluster. You may include more than one additional port mapping by specifying multiple --additional-open-ports arguments.
- --additional-port-mappings
Additional ports to open on the cluster. Mappings should be in the format '[name:]port[/protocol]'. If protocol is not specified, HTTPS will be used. The name portion cannot be included for HTTPS and GRPC protocols; it is required for the others. Valid protocols: (TCP|UDP|HTTPS|GRPC). You may include more than one additional port mapping by specifying multiple --additional-port-mappings arguments.
- --dataset-mount
A mount point for a dataset. Enter in the format of: '<id>:<mount point path>' The 'id' value must be an integer. You may include more than one dataset mountpoint by specifying multiple '--dataset-mount' arguments.
- --workspace-mount
A mount point for a workspace. Enter in the format of: '<id>:<mount point path>:<rw>' The 'rw' value should be 'true' if the mount is read/write, and 'false' if it is read-only. You may include more than one workspace mountpoint by specifying multiple '--workspace-mount' arguments.
- --label
A user/reserved/system label that describes this job. You may define more than one label by specifying multiple '--label' arguments.
- --labels-locked
Labels will not be able to be changed. Default=False
- --multi-node
Only used for jupyterlab cluster types. Determines if the cluster is multi-node or not. Default=False
- --job-order
Order of the job; from 1 to 99.
- --job-priority
Priority of the job; choose from 'LOW', 'NORMAL', or 'HIGH'. Default='NORMAL'
- --min-availability
Minimum replicas that need to be scheduled to start a multi-node job.
- --min-time-slice
Minimum duration (in the format [nD][nH][nM][nS]) the job is expected (not guaranteed) to be in the RUNNING state once scheduled to assure forward progress.
- --options
A custom envronment variable to add to job in the form of a key-pair. A key name must be between 1-63 characters and contain letters, numbers or './-_'. May be used multiple times in the same command.
- --preempt-class
Describes the job class for preemption and scheduling behavior. It must be one of 'RESUMABLE', 'RESTARTABLE', or 'RUNONCE' (default).
- --topology-constraint
Possible choices: any, pack
Specifies a topology constraint for the job. Only available for multi-node jobs. Choices: pack, any.
- --user-secret
Specify secret name for the jab. Format: '<secret[:key_name:alias_name]>'. Multiple secret arguments are allowed. Unless specified all key value pairs will be included in the job. Overrides of the key are available optionally per key-value pair.
Required named arguments¶
- --cluster-type
Possible choices: dask, jupyterlab
The type of cluster: choose from ['dask', 'jupyterlab'].
- --scheduler-instance-type
The instance type of the scheduler
- --nworkers
Number of workers in the cluster
- --container-image
The container image to use
- --data-output-mount-point
The path to where the data output will be mounted
- --cluster-lifetime
The lifetime for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
update¶
Update an existing cluster
ngc base-command quickstart cluster update
[--additional-open-ports ADDITIONAL_OPEN_PORTS]
[--additional-port-mappings ADDITIONAL_PORT_MAPPINGS]
[--cluster-lifetime CLUSTER_LIFETIME]
[--conda-packages CONDA_PACKAGES]
[--container-image CONTAINER_IMAGE]
[--data-output-mount-point DATA_OUTPUT_MOUNT_POINT]
[--dataset-mount DATASET_MOUNT]
[--debug]
[--expiry-duration EXPIRY_DURATION]
[--format_type <fmt>]
[--job-order JOB_ORDER]
[--job-priority JOB_PRIORITY]
[--label LABEL] [--labels-locked]
[--min-availability MIN_AVAILABILITY]
[--min-time-slice MIN_TIME_SLICE]
[--multi-node] [--name NAME]
[--nworkers NWORKERS]
[--options <key:value>]
[--pip-packages PIP_PACKAGES]
[--preempt-class PREEMPT_CLASS]
[--remove-dataset-mounts]
[--remove-workspace-mounts]
[--scheduler-dashboard-address SCHEDULER_DASHBOARD_ADDRESS]
[--scheduler-env-var SCHEDULER_ENV_VAR]
[--scheduler-instance-type SCHEDULER_INSTANCE_TYPE]
[--scheduler-port SCHEDULER_PORT]
[--scheduler-reserved-gpus SCHEDULER_RESERVED_GPUS]
[--scheduler-startup-script SCHEDULER_STARTUP_SCRIPT]
[--system-packages SYSTEM_PACKAGES]
[--topology-constraint {any,pack}]
[--user-secret <secret[:key_name:alias_name]>]
[--worker-dashboard-address WORKER_DASHBOARD_ADDRESS]
[--worker-env-var WORKER_ENV_VAR]
[--worker-instance-type WORKER_INSTANCE_TYPE]
[--worker-reserved-gpus WORKER_RESERVED_GPUS]
[--worker-startup-script WORKER_STARTUP_SCRIPT]
[--workspace-mount WORKSPACE_MOUNT]
[-h]
cluster_id
Positional Arguments¶
- cluster_id
The ID of the cluster
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --name
The name of the cluster
- --scheduler-dashboard-address
The dashboard address for the scheduler. Only used for 'dask' cluster type.
- --scheduler-instance-type
The instance type of the scheduler
- --scheduler-port
The port to use for the scheduler. Only used for 'dask' cluster type.
- --scheduler-startup-script
The startup script for the scheduler
- --scheduler-reserved-gpus
The number of GPUs reserved for the scheduler
- --scheduler-env-var
An environment variable to be set in the scheduler node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--scheduler-env-var' arguments.
- --worker-dashboard-address
The dashboard address for the worker. Only used for 'dask' cluster type.
- --worker-instance-type
The instance type of the worker. Only used for 'dask' cluster type.
- --worker-startup-script
The startup script for the worker. Only used for 'dask' cluster type.
- --worker-reserved-gpus
The number of GPUs reserved for the worker. Only used for 'dask' cluster type.
- --worker-env-var
An environment variable to be set in the worker node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--worker-env-var' arguments. Only used for 'dask' cluster type.
- --system-packages
List of packages to install on the scheduler and worker using 'apt install' or 'yum install' command. (apt or yum command is chosen depending on the flavour of the Linux image in the Container). You may define more than one package by specifying multiple '--system-packages' arguments.
- --conda-packages
List of packages to install on the scheduler and worker using 'conda install' command. You may define more than one package by specifying multiple '--conda-packages' arguments.
- --pip-packages
List of packages to install on the scheduler and worker using 'pip install' command. You may define more than one package by specifying multiple '--pip-packages' arguments.
- --nworkers
Number of workers in the cluster
- --container-image
The container image to use
- --data-output-mount-point
The path to where the data output will be mounted
- --cluster-lifetime
The lifetime for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
- --expiry-duration
The expiry duration for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
- --additional-open-ports
(deprecated; use additional-port-mappings instead) Any additional ports to open for the cluster. You may include more than one additional port mapping by specifying multiple --additional-open-ports arguments.
- --additional-port-mappings
Additional ports to open on the cluster. Mappings should be in the format '[name:]port[/protocol]'. If protocol is not specified, HTTPS will be used. The name portion cannot be included for HTTPS and GRPC protocols; it is required for the others. Valid protocols: (TCP|UDP|HTTPS|GRPC). You may include more than one additional port mapping by specifying multiple --additional-port-mappings arguments.
- --dataset-mount
A mount point for a dataset. Enter in the format of: '<id>:<mount point path>' The 'id' value must be an integer. You may include more than one dataset mountpoint by specifying multiple '--dataset-mount' arguments.
- --workspace-mount
A mount point for a workspace. Enter in the format of: '<id>:<mount point path>:<rw>' The 'rw' value should be 'true' if the mount is read/write, and 'false' if it is read-only. You may include more than one workspace mountpoint by specifying multiple '--workspace-mount' arguments.
- --label
A user/reserved/system label that describes this job. You may define more than one label by specifying multiple '--label' arguments.
- --labels-locked
Labels will not be able to be changed. Default=False
- --multi-node
Is this a multi-node cluster? Default=False
- --job-order
Order of the job; from 1 to 99.
- --job-priority
Priority of the job; choose from 'LOW', 'NORMAL', or 'HIGH'. Default='NORMAL'
- --min-availability
Minimum replicas that need to be scheduled to start a multi-node job.
- --min-time-slice
Minimum duration (in the format [nD][nH][nM][nS]) the job is expected (not guaranteed) to be in the RUNNING state once scheduled to assure forward progress.
- --options
A custom envronment variable to add to job in the form of a key-pair. A key name must be between 1-63 characters and contain letters, numbers or './-_'. May be used multiple times in the same command.
- --preempt-class
Describes the job class for preemption and scheduling behavior. It must be one of 'RESUMABLE', 'RESTARTABLE', or 'RUNONCE' (default).
- --topology-constraint
Possible choices: any, pack
Specifies a topology constraint for the job. Only available for multi-node jobs. Choices: pack, any.
- --user-secret
Specify secret name for the jab. Format: '<secret[:key_name:alias_name]>'. Multiple secret arguments are allowed. Unless specified all key value pairs will be included in the job. Overrides of the key are available optionally per key-value pair.
- --remove-dataset-mounts
Remove any existing dataset mounts for this cluster
- --remove-workspace-mounts
Remove any existing workspace mounts for this cluster
project¶
QuickStart Project Commands
ngc base-command quickstart project [--debug] [--format_type <fmt>] [-h]
{cluster-create,cluster-remove,create,create-template,info,info-template,list,list-templates,remove,remove-template,update,update-template}
...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
project¶
- {cluster-create,cluster-remove,create,create-template,info,info-template,list,list-templates,remove,remove-template,update,update-template}
Possible choices: cluster-create, cluster-remove, create, create-template, info, info-template, list, list-templates, remove, remove-template, update, update-template
Sub-commands¶
list¶
List projects.
ngc base-command quickstart project list [--column <column>] [--debug]
[--format_type <fmt>] [--org-only]
[--owned] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is name[=Name]. Valid columns are ace[=ACE], description[=Description], id[=ID], name[=Name], org[=Org], owner[=Owner], team[=Team]. Use quotes with spaces. Multiple column arguments are allowed.
- --org-only
Don't return clusters created at the team level
- --owned
Only return clusters I own (admin only)
create¶
Create a new project
ngc base-command quickstart project create [--debug] --description DESCRIPTION
[--format_type <fmt>] --name NAME
[--owner OWNER] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --owner
The owner of the project. If not specified, the email for the current user will be used.
Required named arguments¶
- --name
The name of the project
- --description
A desription of the project
update¶
Update a project
ngc base-command quickstart project update [--debug]
[--description DESCRIPTION]
[--format_type <fmt>] [--name NAME]
[-h]
project_id
Positional Arguments¶
- project_id
The ID of the project
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --name
The name of the project
- --description
A desription of the project
info¶
Show information about a project
ngc base-command quickstart project info [--debug] [--format_type <fmt>] [-h]
project_id
Positional Arguments¶
- project_id
The ID of the project
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
remove¶
Delete a project
ngc base-command quickstart project remove [--debug] [--format_type <fmt>]
[-h]
project_id
Positional Arguments¶
- project_id
The ID of the project
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
cluster-create¶
Create a new cluster for the specified project
ngc base-command quickstart project cluster-create
[--additional-open-ports ADDITIONAL_OPEN_PORTS]
[--additional-port-mappings ADDITIONAL_PORT_MAPPINGS]
--cluster-lifetime
CLUSTER_LIFETIME
--cluster-type
{dask,jupyterlab}
[--conda-packages CONDA_PACKAGES]
--container-image
CONTAINER_IMAGE
--data-output-mount-point
DATA_OUTPUT_MOUNT_POINT
[--dataset-mount DATASET_MOUNT]
[--debug]
[--expiry-duration EXPIRY_DURATION]
[--format_type <fmt>]
[--job-order JOB_ORDER]
[--job-priority JOB_PRIORITY]
[--label LABEL]
[--labels-locked]
[--min-availability MIN_AVAILABILITY]
[--min-time-slice MIN_TIME_SLICE]
[--multi-node]
[--name NAME] --nworkers
NWORKERS
[--options <key:value>]
[--pip-packages PIP_PACKAGES]
[--preempt-class PREEMPT_CLASS]
[--scheduler-dashboard-address SCHEDULER_DASHBOARD_ADDRESS]
[--scheduler-env-var SCHEDULER_ENV_VAR]
--scheduler-instance-type
SCHEDULER_INSTANCE_TYPE
[--scheduler-port SCHEDULER_PORT]
[--scheduler-reserved-gpus SCHEDULER_RESERVED_GPUS]
[--scheduler-startup-script SCHEDULER_STARTUP_SCRIPT]
[--system-packages SYSTEM_PACKAGES]
[--topology-constraint {any,pack}]
[--user-secret <secret[:key_name:alias_name]>]
[--worker-dashboard-address WORKER_DASHBOARD_ADDRESS]
[--worker-env-var WORKER_ENV_VAR]
[--worker-instance-type WORKER_INSTANCE_TYPE]
[--worker-reserved-gpus WORKER_RESERVED_GPUS]
[--worker-startup-script WORKER_STARTUP_SCRIPT]
[--workspace-mount WORKSPACE_MOUNT]
[-h]
project_id
Positional Arguments¶
- project_id
The ID of the project
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --name
The name of the cluster
- --scheduler-dashboard-address
The dashboard address for the scheduler. Only used for 'dask' cluster type.
- --scheduler-port
The port to use for the scheduler. Only used for 'dask' cluster type.
- --scheduler-startup-script
The startup script for the scheduler
- --scheduler-reserved-gpus
The number of GPUs reserved for the scheduler
- --scheduler-env-var
An environment variable to be set in the scheduler node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--scheduler-env-var' arguments.
- --worker-dashboard-address
The dashboard address for the worker. Only used for 'dask' cluster type.
- --worker-instance-type
The instance type of the worker. Required for 'dask' cluster type; ignored otherwise.
- --worker-startup-script
The startup script for the worker. Only used for 'dask' cluster type.
- --worker-reserved-gpus
The number of GPUs reserved for the worker. Only used for 'dask' cluster type.
- --worker-env-var
An environment variable to be set in the worker node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--worker-env-var' arguments. Only used for 'dask' cluster type.
- --system-packages
List of packages to install on the scheduler and worker using 'apt install' or 'yum install' command. (apt or yum command is chosen depending on the flavour of the Linux image in the Container). You may define more than one package by specifying multiple '--system-packages' arguments.
- --conda-packages
List of packages to install on the scheduler and worker using 'conda install' command. You may define more than one package by specifying multiple '--conda-packages' arguments.
- --pip-packages
List of packages to install on the scheduler and worker using 'pip install' command. You may define more than one package by specifying multiple '--pip-packages' arguments.
- --expiry-duration
The expiry duration for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
- --additional-open-ports
(deprecated; use additional-port-mappings instead) Any additional ports to open for the cluster. You may include more than one additional port mapping by specifying multiple --additional-open-ports arguments.
- --additional-port-mappings
Additional ports to open on the cluster. Mappings should be in the format '[name:]port[/protocol]'. If protocol is not specified, HTTPS will be used. The name portion cannot be included for HTTPS and GRPC protocols; it is required for the others. Valid protocols: (TCP|UDP|HTTPS|GRPC). You may include more than one additional port mapping by specifying multiple --additional-port-mappings arguments.
- --dataset-mount
A mount point for a dataset. Enter in the format of: '<id>:<mount point path>' The 'id' value must be an integer. You may include more than one dataset mountpoint by specifying multiple '--dataset-mount' arguments.
- --workspace-mount
A mount point for a workspace. Enter in the format of: '<id>:<mount point path>:<rw>' The 'rw' value should be 'true' if the mount is read/write, and 'false' if it is read-only. You may include more than one workspace mountpoint by specifying multiple '--workspace-mount' arguments.
- --label
A user/reserved/system label that describes this job. You may define more than one label by specifying multiple '--label' arguments.
- --labels-locked
Labels will not be able to be changed. Default=False
- --multi-node
Only used for jupyterlab cluster types. Determines if the cluster is multi-node or not. Default=False
- --job-order
Order of the job; from 1 to 99.
- --job-priority
Priority of the job; choose from 'LOW', 'NORMAL', or 'HIGH'. Default='NORMAL'
- --min-availability
Minimum replicas that need to be scheduled to start a multi-node job.
- --min-time-slice
Minimum duration (in the format [nD][nH][nM][nS]) the job is expected (not guaranteed) to be in the RUNNING state once scheduled to assure forward progress.
- --options
A custom envronment variable to add to job in the form of a key-pair. A key name must be between 1-63 characters and contain letters, numbers or './-_'. May be used multiple times in the same command.
- --preempt-class
Describes the job class for preemption and scheduling behavior. It must be one of 'RESUMABLE', 'RESTARTABLE', or 'RUNONCE' (default).
- --topology-constraint
Possible choices: any, pack
Specifies a topology constraint for the job. Only available for multi-node jobs. Choices: pack, any.
- --user-secret
Specify secret name for the jab. Format: '<secret[:key_name:alias_name]>'. Multiple secret arguments are allowed. Unless specified all key value pairs will be included in the job. Overrides of the key are available optionally per key-value pair.
Required named arguments¶
- --cluster-type
Possible choices: dask, jupyterlab
The type of cluster: choose from ['dask', 'jupyterlab'].
- --scheduler-instance-type
The instance type of the scheduler
- --nworkers
Number of workers in the cluster
- --container-image
The container image to use
- --data-output-mount-point
The path to where the data output will be mounted
- --cluster-lifetime
The lifetime for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
cluster-remove¶
Shutdown and delete a cluster from a project
ngc base-command quickstart project cluster-remove --cluster-id CLUSTER_ID
[--debug]
[--format_type <fmt>] [-h]
project_id
Positional Arguments¶
- project_id
The ID of the project
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
Required named arguments¶
- --cluster-id
The ID of the cluster to remove
create-template¶
Create a new project template
ngc base-command quickstart project create-template
[--additional-open-ports ADDITIONAL_OPEN_PORTS]
[--additional-port-mappings ADDITIONAL_PORT_MAPPINGS]
--cluster-lifetime
CLUSTER_LIFETIME
[--cluster-name CLUSTER_NAME]
--cluster-type
{dask,jupyterlab}
[--conda-packages CONDA_PACKAGES]
--container-image
CONTAINER_IMAGE
--data-output-mount-point
DATA_OUTPUT_MOUNT_POINT
[--dataset-mount DATASET_MOUNT]
[--debug] [--default]
--description DESCRIPTION
--display-image-url
DISPLAY_IMAGE_URL
[--expiry-duration EXPIRY_DURATION]
[--format_type <fmt>]
[--job-order JOB_ORDER]
[--job-priority JOB_PRIORITY]
[--label LABEL]
[--labels-locked]
[--min-availability MIN_AVAILABILITY]
[--min-time-slice MIN_TIME_SLICE]
[--multi-node] --name NAME
--nworkers NWORKERS
[--options <key:value>]
[--pip-packages PIP_PACKAGES]
[--preempt-class PREEMPT_CLASS]
[--scheduler-dashboard-address SCHEDULER_DASHBOARD_ADDRESS]
[--scheduler-env-var SCHEDULER_ENV_VAR]
--scheduler-instance-type
SCHEDULER_INSTANCE_TYPE
[--scheduler-port SCHEDULER_PORT]
[--scheduler-reserved-gpus SCHEDULER_RESERVED_GPUS]
[--scheduler-startup-script SCHEDULER_STARTUP_SCRIPT]
[--system-packages SYSTEM_PACKAGES]
[--topology-constraint {any,pack}]
[--user-secret <secret[:key_name:alias_name]>]
[--worker-dashboard-address WORKER_DASHBOARD_ADDRESS]
[--worker-env-var WORKER_ENV_VAR]
[--worker-instance-type WORKER_INSTANCE_TYPE]
[--worker-reserved-gpus WORKER_RESERVED_GPUS]
[--worker-startup-script WORKER_STARTUP_SCRIPT]
[--workspace-mount WORKSPACE_MOUNT]
[-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --cluster-name
The name of the cluster
- --expiry-duration
The expiry duration for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
- --additional-open-ports
(deprecated; use additional-port-mappings instead) Any additional ports to open for the cluster. You may include more than one additional port mapping by specifying multiple --additional-open-ports arguments.
- --additional-port-mappings
Additional ports to open on the cluster. Mappings should be in the format '[name:]port[/protocol]'. If protocol is not specified, HTTPS will be used. The name portion cannot be included for HTTPS and GRPC protocols; it is required for the others. Valid protocols: (TCP|UDP|HTTPS|GRPC). You may include more than one additional port mapping by specifying multiple --additional-port-mappings arguments.
- --scheduler-dashboard-address
The dashboard address for the scheduler. Only used for 'dask' cluster type.
- --scheduler-port
The port to use for the scheduler. Only used for 'dask' cluster type.
- --scheduler-startup-script
The startup script for the scheduler
- --scheduler-reserved-gpus
The number of GPUs reserved for the scheduler
- --scheduler-env-var
An environment variable to be set in the scheduler node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--scheduler-env-var' arguments.
- --worker-dashboard-address
The dashboard address for the worker. Only used for 'dask' cluster type.
- --worker-instance-type
The instance type of the worker. Required for 'dask' cluster type; ignored otherwise.
- --worker-startup-script
The startup script for the worker. Only used for 'dask' cluster type.
- --worker-reserved-gpus
The number of GPUs reserved for the worker. Only used for 'dask' cluster type.
- --worker-env-var
An environment variable to be set in the worker node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--worker-env-var' arguments. Only used for 'dask' cluster type.
- --system-packages
List of packages to install on the scheduler and worker using 'apt install' or 'yum install' command. (apt or yum command is chosen depending on the flavour of the Linux image in the Container). You may define more than one package by specifying multiple '--system-packages' arguments.
- --conda-packages
List of packages to install on the scheduler and worker using 'conda install' command. You may define more than one package by specifying multiple '--conda-packages' arguments.
- --pip-packages
List of packages to install on the scheduler and worker using 'pip install' command. You may define more than one package by specifying multiple '--pip-packages' arguments.
- --dataset-mount
A mount point for a dataset. Enter in the format of: '<id>:<mount point path>' The 'id' value must be an integer. You may include more than one dataset mountpoint by specifying multiple '--dataset-mount' arguments.
- --workspace-mount
A mount point for a workspace. Enter in the format of: '<id>:<mount point path>:<rw>' The 'rw' value should be 'true' if the mount is read/write, and 'false' if it is read-only. You may include more than one workspace mountpoint by specifying multiple '--workspace-mount' arguments.
- --label
A user/reserved/system label that describes this job. You may define more than one label by specifying multiple '--label' arguments.
- --labels-locked
Labels will not be able to be changed. Default=False
- --default
Set this template as the default for the cluster type. Admin only
- --multi-node
Only used for jupyterlab cluster types. Determines if the cluster is multi-node or not. Default=False
- --job-order
Order of the job; from 1 to 99.
- --job-priority
Priority of the job; choose from 'LOW', 'NORMAL', or 'HIGH'. Default='NORMAL'
- --min-availability
Minimum replicas that need to be scheduled to start a multi-node job.
- --min-time-slice
Minimum duration (in the format [nD][nH][nM][nS]) the job is expected (not guaranteed) to be in the RUNNING state once scheduled to assure forward progress.
- --options
A custom envronment variable to add to job in the form of a key-pair. A key name must be between 1-63 characters and contain letters, numbers or './-_'. May be used multiple times in the same command.
- --preempt-class
Describes the job class for preemption and scheduling behavior. It must be one of 'RESUMABLE', 'RESTARTABLE', or 'RUNONCE' (default).
- --topology-constraint
Possible choices: any, pack
Specifies a topology constraint for the job. Only available for multi-node jobs. Choices: pack, any.
- --user-secret
Specify secret name for the jab. Format: '<secret[:key_name:alias_name]>'. Multiple secret arguments are allowed. Unless specified all key value pairs will be included in the job. Overrides of the key are available optionally per key-value pair.
Required named arguments¶
- --name
The name of the project template
- --description
A desription of the project template
- --cluster-type
Possible choices: dask, jupyterlab
The type of cluster: choose from ['dask', 'jupyterlab'].
- --container-image
Container image for the template
- --display-image-url
URL of the image to display for the template
- --nworkers
Number of workers
- --cluster-lifetime
The lifetime for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
- --scheduler-instance-type
The instance type of the scheduler
- --data-output-mount-point
The path to where the data output will be mounted
update-template¶
Modify a project template
ngc base-command quickstart project update-template
[--additional-open-ports ADDITIONAL_OPEN_PORTS]
[--additional-port-mappings ADDITIONAL_PORT_MAPPINGS]
[--cluster-lifetime CLUSTER_LIFETIME]
[--cluster-name CLUSTER_NAME]
[--conda-packages CONDA_PACKAGES]
[--container-image CONTAINER_IMAGE]
[--data-output-mount-point DATA_OUTPUT_MOUNT_POINT]
[--dataset-mount DATASET_MOUNT]
[--debug] [--default]
[--description DESCRIPTION]
[--display-image-url DISPLAY_IMAGE_URL]
[--expiry-duration EXPIRY_DURATION]
[--format_type <fmt>]
[--job-order JOB_ORDER]
[--job-priority JOB_PRIORITY]
[--label LABEL]
[--labels-locked]
[--min-availability MIN_AVAILABILITY]
[--min-time-slice MIN_TIME_SLICE]
[--multi-node]
[--name NAME]
[--nworkers NWORKERS]
[--options <key:value>]
[--pip-packages PIP_PACKAGES]
[--preempt-class PREEMPT_CLASS]
[--remove-dataset-mounts]
[--remove-default]
[--remove-workspace-mounts]
[--scheduler-dashboard-address SCHEDULER_DASHBOARD_ADDRESS]
[--scheduler-env-var SCHEDULER_ENV_VAR]
[--scheduler-instance-type SCHEDULER_INSTANCE_TYPE]
[--scheduler-port SCHEDULER_PORT]
[--scheduler-reserved-gpus SCHEDULER_RESERVED_GPUS]
[--scheduler-startup-script SCHEDULER_STARTUP_SCRIPT]
[--system-packages SYSTEM_PACKAGES]
[--topology-constraint {any,pack}]
[--user-secret <secret[:key_name:alias_name]>]
[--worker-dashboard-address WORKER_DASHBOARD_ADDRESS]
[--worker-env-var WORKER_ENV_VAR]
[--worker-instance-type WORKER_INSTANCE_TYPE]
[--worker-reserved-gpus WORKER_RESERVED_GPUS]
[--worker-startup-script WORKER_STARTUP_SCRIPT]
[--workspace-mount WORKSPACE_MOUNT]
[-h]
template_id
Positional Arguments¶
- template_id
The ID of the template
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --name
The name of the project template
- --cluster-name
The name of the cluster
- --description
A desription of the project template
- --container-image
Container image for the template
- --display-image-url
URL of the image to display for the template
- --nworkers
Number of workers
- --cluster-lifetime
The lifetime for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
- --expiry-duration
The expiry duration for the cluster. The format is <num>X, where 'X' is a single letter representing the unit of time: <d|h|m|s>, for days, hours, minutes, and seconds, respectively.
- --additional-open-ports
(deprecated; use additional-port-mappings instead) Any additional ports to open for the cluster. You may include more than one additional port mapping by specifying multiple --additional-open-ports arguments.
- --additional-port-mappings
Additional ports to open on the cluster. Mappings should be in the format '[name:]port[/protocol]'. If protocol is not specified, HTTPS will be used. The name portion cannot be included for HTTPS and GRPC protocols; it is required for the others. Valid protocols: (TCP|UDP|HTTPS|GRPC). You may include more than one additional port mapping by specifying multiple --additional-port-mappings arguments.
- --scheduler-dashboard-address
The dashboard address for the scheduler. Only used for 'dask' cluster type.
- --scheduler-instance-type
The instance type of the scheduler
- --scheduler-port
The port to use for the scheduler. Only used for 'dask' cluster type.
- --scheduler-startup-script
The startup script for the scheduler
- --scheduler-reserved-gpus
The number of GPUs reserved for the scheduler
- --scheduler-env-var
An environment variable to be set in the scheduler node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--scheduler-env-var' arguments.
- --worker-dashboard-address
The dashboard address for the worker. Only used for 'dask' cluster type.
- --worker-instance-type
The instance type of the worker. Only used for 'dask' cluster type.
- --worker-startup-script
The startup script for the worker. Only used for 'dask' cluster type.
- --worker-reserved-gpus
The number of GPUs reserved for the worker. Only used for 'dask' cluster type.
- --worker-env-var
An environment variable to be set in the worker node. It must be in the format 'var_name:value'. You may define more than one envrionment variable by specifying multiple '--worker-env-var' arguments. Only used for 'dask' cluster type.
- --system-packages
List of packages to install on the scheduler and worker using 'apt install' or 'yum install' command. (apt or yum command is chosen depending on the flavour of the Linux image in the Container). You may define more than one package by specifying multiple '--system-packages' arguments.
- --conda-packages
List of packages to install on the scheduler and worker using 'conda install' command. You may define more than one package by specifying multiple '--conda-packages' arguments.
- --pip-packages
List of packages to install on the scheduler and worker using 'pip install' command. You may define more than one package by specifying multiple '--pip-packages' arguments.
- --data-output-mount-point
The path to where the data output will be mounted
- --dataset-mount
A mount point for a dataset. Enter in the format of: '<id>:<mount point path>' The 'id' value must be an integer. You may include more than one dataset mountpoint by specifying multiple '--dataset-mount' arguments.
- --workspace-mount
A mount point for a workspace. Enter in the format of: '<id>:<mount point path>:<rw>' The 'rw' value should be 'true' if the mount is read/write, and 'false' if it is read-only. You may include more than one workspace mountpoint by specifying multiple '--workspace-mount' arguments.
- --label
A user/reserved/system label that describes this job. You may define more than one label by specifying multiple '--label' arguments.
- --labels-locked
Labels will not be able to be changed. Default=False
- --default
Set this template as the default for the cluster type. Admin only
- --remove-default
Unmark this template as the default for this template's cluster type. Admin only
- --multi-node
Only used for jupyterlab cluster types. Determines if the cluster is multi-node or not. Default=False
- --job-order
Order of the job; from 1 to 99.
- --job-priority
Priority of the job; choose from 'LOW', 'NORMAL', or 'HIGH'. Default='NORMAL'
- --min-availability
Minimum replicas that need to be scheduled to start a multi-node job.
- --min-time-slice
Minimum duration (in the format [nD][nH][nM][nS]) the job is expected (not guaranteed) to be in the RUNNING state once scheduled to assure forward progress.
- --options
A custom envronment variable to add to job in the form of a key-pair. A key name must be between 1-63 characters and contain letters, numbers or './-_'. May be used multiple times in the same command.
- --preempt-class
Describes the job class for preemption and scheduling behavior. It must be one of 'RESUMABLE', 'RESTARTABLE', or 'RUNONCE' (default).
- --topology-constraint
Possible choices: any, pack
Specifies a topology constraint for the job. Only available for multi-node jobs. Choices: pack, any.
- --user-secret
Specify secret name for the jab. Format: '<secret[:key_name:alias_name]>'. Multiple secret arguments are allowed. Unless specified all key value pairs will be included in the job. Overrides of the key are available optionally per key-value pair.
- --remove-dataset-mounts
Remove any existing dataset mounts for this cluster
- --remove-workspace-mounts
Remove any existing workspace mounts for this cluster
list-templates¶
List project templates.
ngc base-command quickstart project list-templates [--column <column>]
[--debug] [--default-only]
[--format_type <fmt>]
[--template-type TEMPLATE_TYPE]
[-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is name[=Name]. Valid columns are id[=ID], name[=Name], description[=Description], display-image-url[="Display Image"]. Use quotes with spaces. Multiple column arguments are allowed.
- --default-only
Only list default template
- --template-type
Type of template to show. Choices: dask, jupyterlab. Default='dask'
remove-template¶
Delete a project template
ngc base-command quickstart project remove-template [--debug]
[--format_type <fmt>] [-h]
template_id
Positional Arguments¶
- template_id
The ID of the template
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
info-template¶
Get information about a project template
ngc base-command quickstart project info-template [--debug]
[--format_type <fmt>] [-h]
template_id
Positional Arguments¶
- template_id
The ID of the template
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
resource¶
Resource Commands
ngc base-command resource [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
resource¶
Possible choices: create, info, list, remove, telemetry, update
Sub-commands¶
list¶
List resources. Specify ace to list child pools, or 'no-ace' to list across aces.
ngc base-command resource list [--column <column>] [--debug]
[--format_type <fmt>] [--root-pool]
[--user-id <user_id>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --user-id
Specify user id, use user-defaults for default pool.
- --column
Specify output column as column[=header], header is optional, default is id[=Id]. Valid columns are id[=Id], poolType[=Type], description[=Description], resourceTypeName[="Resource Type"]. Use quotes with spaces. Multiple column arguments are allowed.
- --root-pool
Use to list root pools instead.
info¶
Resource information.
ngc base-command resource info [--debug] [--format_type <fmt>] [--root-pool]
[--user-id <user_id>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --user-id
Specify user id, use user-defaults for default pool.
- --root-pool
Use to get root pool info instead.
create¶
Create Pool.
ngc base-command resource create [--allocation <allocation>] [--debug]
[--default <default>]
[--description <description>]
[--format_type <fmt>] [--user-id <user_id>]
[--version <version>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --user-id
Specify user id, use user-defaults for default pool.
- --description
Specify pool description.
- --version
Specify pool version.
- --allocation
Specify resource allocations for the pool. Format <type>:<limit>:<share>:<priority>. Either allocation or default is required.
- --default
Specify resource defaults for the pool, required for creating team pool. Format <type>:<limit>:<share>:<priority>. Either allocation or default is required.
update¶
Update Pool.
ngc base-command resource update [--add-allocation <add_allocation>] [--debug]
[--description <description>]
[--format_type <fmt>]
[--remove-allocation <remove_allocation>]
[--update-allocation <update_allocation>]
[--user-id <user_id>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --description
Specify pool description.
- --update-allocation
Specify resource allocations to update. Format <type>:<limit>:<share>:<priority>.
- --add-allocation
Specify resource allocations to add to the pool. Format <type>:<limit>:<share>:<priority>.
- --remove-allocation
Specify resource allocations to remove from the pool. Format <resource_type>.
- --user-id
Specify user id, use user-defaults for default pool.
remove¶
Remove Pool.
ngc base-command resource remove [--debug] [--format_type <fmt>]
[--user-id <user_id>] [-h] [-y]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --user-id
Specify user id, use user-defaults for default pool.
- -y, --yes
Automatically say yes to all interactive questions.
telemetry¶
Resource telemetry.
ngc base-command resource telemetry [--debug] [--end-time <end_time>]
[--format_type <fmt>]
[--interval-time <interval_time>]
[--interval-unit <interval_unit>]
--resource-type <resource_type>
[--root-pool] [--start-time <start-time>]
[--telemetry-type <telemetry_type>]
[--user-id <user_id>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --user-id
Specify user id, use user-defaults for default pool.
- --telemetry-type
Possible choices: ACTIVE_FAIR_SHARE, FAIR_SHARE, POOL_CAPACITY, POOL_LIMIT, QUEUED_JOBS, RESOURCE_USAGE, RESOURCE_UTILIZATION, RUNNING_JOBS, im_resource_manager_active_rcrs_per_pool_total, im_resource_manager_num_resources_consumed_total, im_resource_manager_num_resources_needed_total, im_resource_manager_pending_rcrs_per_pool_total, im_resource_manager_pool_limit_total, im_resource_manager_pool_reservation_total, im_resource_manager_pool_share_total
Specify type for telemetry. Options ['ACTIVE_FAIR_SHARE', 'FAIR_SHARE', 'POOL_CAPACITY', 'POOL_LIMIT', 'QUEUED_JOBS', 'RESOURCE_USAGE', 'RESOURCE_UTILIZATION', 'RUNNING_JOBS', 'im_resource_manager_active_rcrs_per_pool_total', 'im_resource_manager_num_resources_consumed_total', 'im_resource_manager_num_resources_needed_total', 'im_resource_manager_pending_rcrs_per_pool_total', 'im_resource_manager_pool_limit_total', 'im_resource_manager_pool_reservation_total', 'im_resource_manager_pool_share_total']
- --end-time
Specifies the end time for statistics. Format: [yyyy-MM-dd::HH:mm:ss]. Default: now
- --start-time
Specifies the start time for statistics. Format: [yyyy-MM-dd::HH:mm:ss].
- --interval-unit
Possible choices: HOUR, MINUTE, SECOND
Data collection interval unit. Options: ['HOUR', 'MINUTE', 'SECOND']. Default: MINUTE
- --interval-time
Data collection interval time value. Default: 1
- --root-pool
Use to get root pool info instead.
Required named arguments¶
- --resource-type
Filter data by resource type.
result¶
Job Result Commands
ngc base-command result [--debug] [--format_type <fmt>] [-h] ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
result¶
Possible choices: download, info, remove
Sub-commands¶
info¶
List result file(s) generated by the successful execution of a job.
ngc base-command result info [--debug] [--files] [--format_type <fmt>] [-h]
<<job id>[:replica_id]>
Positional Arguments¶
- <<job id>[:replica_id]>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --files
Include list of files with result details. The maximum number of files shown is 1000.
download¶
Download results by job ID.
ngc base-command result download [--debug] [--dest <path>] [--dir <wildcard>]
[--dry-run] [--exclude <wildcard>]
[--file <wildcard>] [--format_type <fmt>]
[--resume <resume>] [--zip] [-h]
<job_id>[:replica_id]>
Positional Arguments¶
- <job_id>[:replica_id]>
Job ID
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Target destination for the downloaded file(s). Default: .
- --file
File name(s) to be downloaded from the resultset. This flag can be supplied multiple times, and supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..). This flag only filters based on the name of the file, not the path. The maximum number of files downloaded per filter is 1000.
- --dir
Directory or directories to be downloaded from the resultset. This flag supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..). The maximum number of files downloaded per filter is 1000.
- --exclude
The file(s) or directory (directories) to be excluded from the download. This flag can be supplied multiple times, and supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..)
- --zip
Download the entire result directory as a zip file.
- --dry-run
List total size of the download without performing the download.
- --resume
Resume the download for the result. Specify the file name saved by the download. Files will be downloaded to the directory of the file name.
remove¶
Remove a result for given job ID(s).
ngc base-command result remove [--debug] [--dry-run] [--format_type <fmt>]
[--job-name <wildcard>] [--status <status>]
[-h] [-y]
<jobid|jobidrange|jobidlist|jobidpattern>
Positional Arguments¶
- <jobid|jobidrange|jobidlist|jobidpattern>
Job ID(s). Valid examples: '1-5', '1,10-15', '*', '1?'
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --status
Possible choices: CANCELED, FAILED, FAILED_RUN_LIMIT_EXCEEDED, FINISHED_SUCCESS, IM_INTERNAL_ERROR, INFINITY_POOL_MISSING, KILLED_BY_ADMIN, KILLED_BY_SYSTEM, KILLED_BY_USER, PENDING_TERMINATION, PREEMPTED, PREEMPTED_BY_ADMIN, RESOURCE_CONSUMPTION_REQUEST_IN_PROGRESS, RESOURCE_GRANTED, RESOURCE_GRANT_DENIED, RESOURCE_LIMIT_EXCEEDED, RESOURCE_RELEASED, RUNNING, TASK_LOST, UNKNOWN
Filter jobs matching input status. Multiple --status flags will OR together. Options: ['CANCELED', 'FAILED', 'FAILED_RUN_LIMIT_EXCEEDED', 'FINISHED_SUCCESS', 'IM_INTERNAL_ERROR', 'INFINITY_POOL_MISSING', 'KILLED_BY_ADMIN', 'KILLED_BY_SYSTEM', 'KILLED_BY_USER', 'PENDING_TERMINATION', 'PREEMPTED', 'PREEMPTED_BY_ADMIN', 'RESOURCE_CONSUMPTION_REQUEST_IN_PROGRESS', 'RESOURCE_GRANTED', 'RESOURCE_GRANT_DENIED', 'RESOURCE_LIMIT_EXCEEDED', 'RESOURCE_RELEASED', 'RUNNING', 'TASK_LOST', 'UNKNOWN']
- -y, --yes
Automatically say yes to all interactive questions.
- --dry-run
List matching input set without any action.
- --job-name
Filter jobs with names matching specified pattern. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..).
Examples¶
How to list ACE?¶
$ ngc base-command ace list
+-----------------------+-------+---------------------------------------+-----------------------------+
| ACE | Id | Description | Instances |
+-----------------------+-------+---------------------------------------+-----------------------------+
| nvidia-exmpl-1 | 63915 | For testing group A | nvidia.exmpl.a2, |
| | | | nvidia.exmpl.a6 |
| nvidia-exmpl-2 | 56392 | For testing group B | nvidia.exmpl.b7, |
| | | | nvidia.exmpl.b5 |
+-----------------------+-------+---------------------------------------+-----------------------------+
How to list ACE using column arguments?¶
$ ngc base-command ace list --column id
+-----------------------+-------+
| ACE | Id |
+-----------------------+-------+
| nvidia-exmpl-1 | 63915 |
| nvidia-exmpl-2 | 56392 |
+-----------------------+-------+
How to get ACE details?¶
$ ngc base-command ace info nvidia-exmpl-1
----------------------------------------------------
ACE Information
Name: nvidia-exmpl-1
Id: 63915
Type: PUBLIC
Created Date: 2018-03-04 05:01:00 UTC
Created By:
Description: For testing group A
Auto Configuration Enabled: False
Provider: NGN
Storage Service Url: https://nvidia.com
Proxy Service Url: https://nvidia.com
Instances:
Name: nvidia.exmpl.a2 GPUs: 1 GPU Mem: 16 GB GPU Power: 160 W CPUs: 8 System Mem: 50 GB
Name: nvidia.exmpl.a6 GPUs: 2 GPU Mem: 16 GB GPU Power: 160 W CPUs: 16 System Mem: 100 GB
----------------------------------------------------
How to list dataset?¶
$ ngc base-command dataset list
+------------+--------------+------------------+--------+----------+-----------------+----------------------------+
| Dataset Id | Dataset Name | Description | ACE Id | Size | Status | Created Date |
+------------+--------------+------------------+--------+----------+-----------------+----------------------------+
| 3067 | mnist | mnist is awesome | 53 | 11.06 MB | UPLOAD_COMPLETE | 2017-12-02 22:21:38 UTC +0 |
+------------+--------------+------------------+--------+----------+-----------------+----------------------------+
How to list dataset using column arguments?¶
$ ngc base-command dataset list --column id --column name --column size
+-----------------------+--------------+------------------+----------+
| Id | Integer Id | Name | Size |
+-----------------------+--------------+------------------+----------+
| oSjpnu6GtEXRQZCqqDp31 | 3067 | mnist is awesome | 11.06 MB |
+-----------------------+--------------+------------------+----------+
How to upload dataset?¶
$ ngc base-command dataset upload --ace nv-exmpl-1 --desc "mnist is awesome" --source c:\dataset\mnist.zip mnist
Parsing list of files...
Number of files to be uploaded: 1
Creating dataset...
Dataset created id: 4597
Upload started: mnist.zip
2017-12-19 10:28:57:963000 Upload completed: mnist.zip Time taken: 3.28 seconds File size: 11600578 B
Total number of files uploaded 1/1
Dataset: 4597 Name: mnist Upload Completed
Dataset local path: c:\dataset\mnist.zip
Files uploaded: 1
Total Bytes transferred: 11600578 B
Started at: 2017-12-19 10:28:52.738000
Completed at: 2017-12-19 10:28:57.970000
Duration taken: 5.232 seconds
NOTE: It will take some time for dataset to be available for download.
How to view dataset info?¶
$ ngc base-command dataset info 4597
----------------------------------------------------
Dataset Information
Id: 4597
Name: mnist
Created By: FooBar
Email: foobar@nvidia.com
ACE: nv-exmpl-1
Size: 11.6 MB
Total Files: 1
Status: UPLOAD_COMPLETE
Description: mnist is awesome
Owned: No
Shared: Private
Files:
/mnist.zip
----------------------------------------------------
How to download dataset?¶
$ ngc base-command dataset download 9999
1 files to download, total size - 13 B
Downloaded 0 B, 0 files in 2s Download speed: 0 B/s
Downloaded 0 B, 0 files in 3s Download speed: 0 B/s
**********************************************************************
Dataset: 9999 Download status: Completed.
Downloaded local path: C:\Users\Admin\ngc-cli\9999
Total files downloaded: 1
Total downloaded size: 13 B
Started at: 2018-01-01 10:00:26.756000
Completed at: 2018-01-01 10:00:31.720000
Duration taken: 4s seconds
**********************************************************************
How to remove dataset?¶
$ ngc base-command dataset remove 4597
How to convert result to dataset?¶
$ ngc base-command dataset convert --from-result 9999 dataset-name
How to submit a job?¶
$ ngc base-command job run --ace nv-exmpl-1 --instance ngcv1 --name "Test Run"
--image "nvidia/pytorch:17.11" --datasetid 5586:/data
--result /result --command "/bin/ls -aFl /data"
Job created.
----------------------------------------------------
Job Information
Id: 813581
Name: Test Run
Job Type: BATCH
Created At: 2019-09-09 20:28:32 UTC
Submitted By: <your name>
Job Container Information
Docker Image URL: nvidia/pytorch:17.11
Job Commands
Command: /bin/ls -aFl /data
Dockerfile Image Entrypoint: False
Datasets, Workspaces and Results
Dataset ID: 5586
Dataset Mount Point: /data
Result Mount Point: /result
Job Resources
Instance Type: ngcv1
Instance Details: 1 GPU, 8 CPU, 50 GB System Memory
ACE: nv-exmpl-1
Cluster: prd0-257-sjc3.nvk8s.com
Job Status
Status: CREATED
Preempt Class: RUNONCE
----------------------------------------------------
How to submit job using json file?¶
1. First upload the dataset to ACE.
$ ngc base-command dataset upload --ace nv-exmpl-1 --desc "mnist is awesome" --source c:\dataset\mnist.zip mnist
Parsing list of files...
Number of files to be uploaded: 1
Creating dataset...
Dataset created id: 4597
Upload started: mnist.zip
2017-12-19 10:28:57:963000 Upload completed: mnist.zip Time taken: 3.28 seconds File size: 11600578 B
Total number of files uploaded 1/1
Dataset: 4597 Name: mnist Upload Completed
Dataset local path: c:\dataset\mnist.zip
Files uploaded: 1
Total Bytes transferred: 11600578 B
Started at: 2017-12-19 10:28:52.738000
Duration taken: 5.232 seconds
Completed at: 2017-12-19 10:28:57.970000
NOTE: It will take some time for dataset to be available for download.
2. Create a tensorflow-mnist.json file with following json object. Make sure you use the dataset id generated
by uploading the dataset.
{
"dockerImageName": "nvidia/tensorflow:17.10",
"aceName": "nv-exmpl-1",
"name": "tensorflow-mnist",
"command": "cp -r /data /tmp/; cd /tmp/data; unzip -j mnist.zip; cd /src/TensorFlow-Examples/examples/3_NeuralNetworks; cp /tmp/data/input_data.py .; python /src/TensorFlow-Examples/examples/3_NeuralNetworks/convolutional_network.py",
"datasetMounts": [
{
"containerMountPoint": "/data",
"id": 4597
}
],
"resultContainerMountPoint": "/result",
"aceInstance": "ngcv1"
}
3. Submit job using json file.
$ ngc base-command job run -f tensorflow-mnist.json
How to submit job using json file and arguments?¶
Note: Single-use arguments will override and multiple-use arguments will be appended to corresponding json values.
Submit job with a new name and instance overriding the values from the json file.
$ ngc base-command job run -f tensorflow-mnist.json --name tensorflow-mnist-2 --instance dgx1v.16g.2.norm
--------------------------------------------------
Job Information
Id: 1181406
Name: tensorflow-mnist-2
Number of Replicas: 1
Job Type: BATCH
Submitted By: <your name>
Job Container Information
Docker Image URL: nvidia/tensorflow:17.10
Job Commands
Command: cp -r /data /tmp/; cd /tmp/data; unzip -j mnist.zip; cd /src/TensorFlow-Examples/examples/3_NeuralNetworks; cp /tmp/data/input_data.py .; python /src/TensorFlow-Examples/examples/3_NeuralNetworks/convolutional_network.py
Dockerfile Image Entrypoint: False
Datasets, Workspaces and Results
Dataset ID: 5586
Dataset Mount Point: /data
Prepopulated: No
Result Mount Point: /result
Job Resources
Instance Type: dgx1v.16g.2.norm
Instance Details: 2 GPU, 16 CPU, 100 GB System Memory
ACE: nv-exmpl-1
Cluster: prd0-257-sjc3.nvk8s.com
Team: ngc
Job Status
Created at: 2020-05-15 19:59:52 UTC
Status: CREATED
Preempt Class: RUNONCE
--------------------------------------------------
How to kill job?¶
$ ngc base-command job kill 19556
Submitted kill request for job Id: 19556
How to list jobs?¶
$ ngc base-command job list
+--------+-------------+-----------+------------------+----------+--------------------------+
| Id | Name | Team |Status | Duration | Status Details |
+--------+----- -------+-----------|------------------+----------+--------------------------+
| 123456 | job-name-4 | | FINISHED_SUCCESS | 0:00:30 | |
| 123455 | job-name-3 | | FINISHED_SUCCESS | 0:00:00 | |
| 314169 | job-name-2 | your_team | KILLED_BY_USER | - | (-1): Requested by user. |
| 314172 | job-name-1 | my+team | FAILED | 0:00:01 | (100): Container exited |
| | | | | | with status 1. |
+--------+-------------+-----------+------------------+----------+--------------------------+
How to list jobs using column arguments?¶
$ ngc base-command job list --column name --column status
+--------+------------+------------------+
| Id | Name |Status |
+--------+----- ------|------------------+
| 123456 | job-name-4 | FINISHED_SUCCESS |
| 123455 | job-name-3 | FINISHED_SUCCESS |
| 314169 | job-name-2 | KILLED_BY_USER |
| 314172 | job-name-1 | FAILED |
+--------+------------+------------------+
How to view job info?¶
$ ngc base-command job info 710135
--------------------------------------------------
Job Information
Id: 710135
Name: TestJobLumetta
Job Type: BATCH
Created At: 2019-06-20 20:38:35 UTC
Submitted By: <your name>
Job Container Information
Docker Image URL: nvidia/pytorch:17.11
Container name: c983092fb5755e946177f82bb01e2eb02021b9285697cd9cf06e8da0e8c382a0
Job Commands
Dockerfile Image Entrypoint: False
Datasets, Workspaces and Results
Result Mount Point: /result
Job Resources
Instance Type: ngcv1
Instance Details: 1 GPU, 8 CPU, 50 GB System Memory
ACE: Staging
Cluster: stg-sjc3.nonprod-nvkong.com
Job Status
Started at: 2019-06-20 20:41:12 UTC
Ended at: 2019-06-20 20:41:12 UTC
Status: FINISHED_SUCCESS
Status Type: OK
Preempt Class: RUNONCE
--------------------------------------------------
How to attach to Docker container?¶
Note: Specified job must be running
$ ngc base-command job attach 15472
How to exec to Docker container?¶
Note: Specified job must be running
$ ngc base-command job exec 15472
How to generate a job with telemetry data?¶
Note: Specified image must have metrics turned on, or there will be no output for the telemetry command.
$ ngc base-command job run --name "test-run" --preempt RUNONCE --ace nv-exmpl-1
--instance dgx1v.16g.2.norm --commandline "jupyter lab --ip=0.0.0.
--allow-root --no-browser --NotebookApp.token=''
--notebook-dir= --NotebookApp.allow_origin='*' & date;
export TF_ENABLE_AUTO_MIXED_PRECISION=1;
cd /mnt/democode/nvidia/tensorflow-19.05-py3/workspace/nvidia-examples/cnn;
mpiexec --allow-run-as-root -np 1 python resnet.py --layers=50
--data_dir=/data/imagenet --log_dir=/result; " --result /result
--image "nvidia/tensorflow:19.11-tf1-py3" --org nvexmpl --team onboarding
--datasetid 9382:/data/imagenet
--workspace wWL-MpYfSkmWcSjvZFwYwQ:/mnt/democode:RO --port 8888
Job created. (2-GPU configuration: "--instance dgx1v.16g.2.norm")
-------------------------------------------------------------------------------------------
Job Information
Id: 1120624
Name: test-run
Number of Replicas: 1
Job Type: BATCH
Created At: 2020-04-07 05:38:38 UTC
Submitted By: <your name>
Job Container Information
Docker Image URL: nvidia/tensorflow:19.11-tf1-py3
Job Commands
Command: jupyter lab --ip=0.0.0.0 --allow-root --no-browser --NotebookApp.token=''
--notebook-dir=/ --NotebookApp.allow_origin='*' & date;
export TF_ENABLE_AUTO_MIXED_PRECISION=1;
cd /mnt/democode/nvidia/tensorflow-19.05-py3/workspace/nvidia-examples/cnn;
mpiexec --allow-run-as-root -np 1 python resnet.py --layers=50
--data_dir=/data/imagenet --log_dir=/result;
Dockerfile Image Entrypoint: False
Datasets, Workspaces and Results
Dataset ID: 9382
Dataset Mount Point: /data/imagenet
Prepopulated: No
Workspace ID: wWL-MpYfSkmWcSjvZFwYwQ
Workspace Name: sn-code
Workspace Mount Point: /mnt/democode
Workspace Mount Mode: RO
Result Mount Point: /result
Job Resources
Instance Type: dgx1v.16g.2.norm
Instance Details: 2 GPU, 8 CPU, 50 GB System Memory
ACE: nv-exmpl-1
Cluster: prd0-257-sjc3.nvk8s.com
Team: onboarding
Job Status
Status: CREATED
Preempt Class: RUNONCE
-------------------------------------------------------------------------------------------
How to get telemetry arguments information?¶
$ ngc base-command job telemetry --help
usage: ngc batch telemetry [--ace <name>] [--debug] [--format_type <fmt>]
[--interval-time <t>] [--interval-unit <u>]
[--org <name>] [--statistics <form>]
[--team <name>] [--type <type>] [-h]
<<job id>[:replica_id]>
List telemetry data for the given job.
positional arguments:
<<job id>[:replica_id]>
Job ID
optional arguments:
-h, --help Show this help message and exit.
--ace <name> Specify the ACE name. Use "--ace no-ace" to override
other sources and specify no ACE. Default: current
configuration
--debug Enable debug mode.
--format_type <fmt> Specify the output format type. Supported formats are:
ascii, csv, json. Only commands that produce tabular
data support csv format. Default: ascii
--interval-time <t> Data collection interval time value. Default: 1
--interval-unit <u> Data collection interval unit. Options: HOUR, MINUTE,
SECOND. Default: MINUTE
--org <name> Specify the organization name. Use "--org no-org" to
override other sources and specify no org. Default:
current configuration
--statistics <form> Statistical form of the data to report. Options: MAX,
MEAN, MIN. Default: MEAN
--team <name> Specify the team name. Use "--team no-team" to
override other sources and specify no team. Default:
current configuration
--type <type> A telemetry type to report. Options:
APPLICATION_TELEMETRY, CPU_UTILIZATION, GPU_FB_USED,
GPU_FI_PROF_DRAM_ACTIVE, GPU_FI_PROF_PCIE_RX_BYTES,
GPU_FI_PROF_PCIE_TX_BYTES,
GPU_FI_PROF_PIPE_TENSOR_ACTIVE,
GPU_NVLINK_BANDWIDTH_TOTAL, GPU_POWER_USAGE,
GPU_UTILIZATION, MEM_UTILIZATION. Default: None
How to view default telemetry?¶
Note: This case shows a 2-GPU configuration (--instance dgx1v.16g.2.norm)
$ ngc base-command job telemetry --interval-time 90 --interval-unit HOUR 1120624
+-----------------------------------+----------------------+-----------------------+
| Name | Time | Measurement |
+-----------------------------------+----------------------+-----------------------+
| ngcjob_appmetrics_job_rate | 2020-05-04T06:00:00Z | 351.479144517938 |
| ngcjob_appmetrics_job_rate | 2020-05-08T00:00:00Z | 372.21862476360195 |
| ngcjob_appmetrics_learn_rate | 2020-05-04T06:00:00Z | 0.7806435847527863 |
| ngcjob_appmetrics_learn_rate | 2020-05-08T00:00:00Z | 0.019446378715977816 |
| ngcjob_appmetrics_num_epochs | 2020-05-04T06:00:00Z | 38.453847984087396 |
| ngcjob_appmetrics_num_epochs | 2020-05-08T00:00:00Z | 81.39723414001247 |
| ngcjob_appmetrics_loss | 2020-05-04T06:00:00Z | 2.6386090555393493 |
| ngcjob_appmetrics_loss | 2020-05-08T00:00:00Z | 1.356071743548495 |
| ngcjob_appmetrics_total_loss | 2020-05-04T06:00:00Z | 4.068851139706283 |
| ngcjob_appmetrics_total_loss | 2020-05-08T00:00:00Z | 1.9573122789348445 |
| GPU_UTILIZATION | 2020-05-04T06:00:00Z | 18.253698644530818 |
| GPU_UTILIZATION | 2020-05-08T00:00:00Z | 19.78424418604651 |
| GPU_UTILIZATION_gpu_5 | 2020-05-04T06:00:00Z | 36.45377190221379 |
| GPU_UTILIZATION_gpu_5 | 2020-05-08T00:00:00Z | 39.70700116686115 |
| GPU_UTILIZATION_gpu_6 | 2020-05-04T06:00:00Z | 0.0 |
| GPU_UTILIZATION_gpu_6 | 2020-05-08T00:00:00Z | 0.0 |
| GPU_FI_PROF_PIPE_TENSOR_ACTIVE | 2020-05-04T06:00:00Z | 5.0732140045141945 |
| GPU_FI_PROF_PIPE_TENSOR_ACTIVE | 2020-05-08T00:00:00Z | 5.531902552204171 |
| GPU_FI_PROF_PIPE_TENSOR_ACTIVE_gp | 2020-05-04T06:00:00Z | 10.155490175475455 |
| u_5 | | |
| GPU_FI_PROF_PIPE_TENSOR_ACTIVE_gp | 2020-05-08T00:00:00Z | 11.076655052264798 |
| u_5 | | |
| GPU_FI_PROF_PIPE_TENSOR_ACTIVE_gp | 2020-05-04T06:00:00Z | 0.0 |
| u_6 | | |
| GPU_FI_PROF_PIPE_TENSOR_ACTIVE_gp | 2020-05-08T00:00:00Z | 0.0 |
| u_6 | | |
| GPU_FI_PROF_DRAM_ACTIVE | 2020-05-04T06:00:00Z | 8.480215505913197 |
| GPU_FI_PROF_DRAM_ACTIVE | 2020-05-08T00:00:00Z | 9.223577235772378 |
| GPU_FI_PROF_DRAM_ACTIVE_gpu_5 | 2020-05-04T06:00:00Z | 16.963105877404956 |
| GPU_FI_PROF_DRAM_ACTIVE_gpu_5 | 2020-05-08T00:00:00Z | 18.46860465116283 |
| GPU_FI_PROF_DRAM_ACTIVE_gpu_6 | 2020-05-04T06:00:00Z | 0.0 |
| GPU_FI_PROF_DRAM_ACTIVE_gpu_6 | 2020-05-08T00:00:00Z | 0.0 |
| GPU_POWER_USAGE | 2020-05-04T06:00:00Z | 70.64337745072271 |
| GPU_POWER_USAGE | 2020-05-08T00:00:00Z | 72.78912856311828 |
| GPU_POWER_USAGE_gpu_5 | 2020-05-04T06:00:00Z | 95.24901714526119 |
| GPU_POWER_USAGE_gpu_5 | 2020-05-08T00:00:00Z | 99.43907317073177 |
| GPU_POWER_USAGE_gpu_6 | 2020-05-04T06:00:00Z | 46.06617461651623 |
| GPU_POWER_USAGE_gpu_6 | 2020-05-08T00:00:00Z | 46.04600233100266 |
| GPU_FB_USED | 2020-05-04T06:00:00Z | 6253.323733445449 |
| GPU_FB_USED | 2020-05-08T00:00:00Z | 6276.2501454333915 |
| GPU_FB_USED_gpu_5 | 2020-05-04T06:00:00Z | 12359.62638056169 |
| GPU_FB_USED_gpu_5 | 2020-05-08T00:00:00Z | 12361.0 |
| GPU_FB_USED_gpu_6 | 2020-05-04T06:00:00Z | 156.0 |
| GPU_FB_USED_gpu_6 | 2020-05-08T00:00:00Z | 156.0 |
| GPU_FI_PROF_PCIE_RX_BYTES | 2020-05-04T06:00:00Z | 5.681953140065174E8 |
| GPU_FI_PROF_PCIE_RX_BYTES | 2020-05-08T00:00:00Z | 6.240817377829639E8 |
| GPU_FI_PROF_PCIE_RX_BYTES_gpu_5 | 2020-05-04T06:00:00Z | 1.1338959633084602E9 |
| GPU_FI_PROF_PCIE_RX_BYTES_gpu_5 | 2020-05-08T00:00:00Z | 1.242978021476135E9 |
| GPU_FI_PROF_PCIE_RX_BYTES_gpu_6 | 2020-05-04T06:00:00Z | 2256750.4371780045 |
| GPU_FI_PROF_PCIE_RX_BYTES_gpu_6 | 2020-05-08T00:00:00Z | 2290032.879532164 |
| GPU_FI_PROF_PCIE_TX_BYTES | 2020-05-04T06:00:00Z | 7.030049576372905E7 |
| GPU_FI_PROF_PCIE_TX_BYTES | 2020-05-08T00:00:00Z | 7.68041546759744E7 |
| GPU_FI_PROF_PCIE_TX_BYTES_gpu_5 | 2020-05-04T06:00:00Z | 1.3947316406226337E8 |
| GPU_FI_PROF_PCIE_TX_BYTES_gpu_5 | 2020-05-08T00:00:00Z | 1.5263490327272728E8 |
| GPU_FI_PROF_PCIE_TX_BYTES_gpu_6 | 2020-05-04T06:00:00Z | 1222276.019745825 |
| GPU_FI_PROF_PCIE_TX_BYTES_gpu_6 | 2020-05-08T00:00:00Z | 1237624.7154471544 |
| GPU_NVLINK_BANDWIDTH_TOTAL | 2020-05-04T06:00:00Z | 0.0 |
| GPU_NVLINK_BANDWIDTH_TOTAL | 2020-05-08T00:00:00Z | 0.0 |
| GPU_NVLINK_BANDWIDTH_TOTAL_gpu_5 | 2020-05-04T06:00:00Z | 0.0 |
| GPU_NVLINK_BANDWIDTH_TOTAL_gpu_5 | 2020-05-08T00:00:00Z | 0.0 |
| GPU_NVLINK_BANDWIDTH_TOTAL_gpu_6 | 2020-05-04T06:00:00Z | 0.0 |
| GPU_NVLINK_BANDWIDTH_TOTAL_gpu_6 | 2020-05-08T00:00:00Z | 0.0 |
| CPU_UTILIZATION | 2020-05-08T00:00:00Z | 0.8535945500785372 |
| MEM_UTILIZATION | 2020-05-04T06:00:00Z | 1.0148062796099368E11 |
| MEM_UTILIZATION | 2020-05-08T00:00:00Z | 9.146897689060422E10 |
+-----------------------------------+----------------------+-----------------------+
How to select the needed telemetry types?¶
$ ngc base-command job telemetry --interval-time 90 --interval-unit HOUR --type GPU_POWER_USAGE
--type APPLICATION_TELEMETRY 1120624
+-----------------------------------+----------------------+-----------------------+
| Name | Time | Measurement |
+-----------------------------------+----------------------+-----------------------+
| GPU_UTILIZATION | 2020-05-04T06:00:00Z | 18.253698644530818 |
| GPU_UTILIZATION | 2020-05-08T00:00:00Z | 19.78424418604651 |
| GPU_UTILIZATION_gpu_5 | 2020-05-04T06:00:00Z | 36.45377190221379 |
| GPU_UTILIZATION_gpu_5 | 2020-05-08T00:00:00Z | 39.70700116686115 |
| GPU_UTILIZATION_gpu_6 | 2020-05-04T06:00:00Z | 0.0 |
| GPU_UTILIZATION_gpu_6 | 2020-05-08T00:00:00Z | 0.0 |
| ngcjob_appmetrics_job_rate | 2020-05-04T06:00:00Z | 351.479144517938 |
| ngcjob_appmetrics_job_rate | 2020-05-08T00:00:00Z | 372.21862476360195 |
| ngcjob_appmetrics_learn_rate | 2020-05-04T06:00:00Z | 0.7806435847527863 |
| ngcjob_appmetrics_learn_rate | 2020-05-08T00:00:00Z | 0.019446378715977816 |
| ngcjob_appmetrics_num_epochs | 2020-05-04T06:00:00Z | 38.453847984087396 |
| ngcjob_appmetrics_num_epochs | 2020-05-08T00:00:00Z | 81.39723414001247 |
| ngcjob_appmetrics_loss | 2020-05-04T06:00:00Z | 2.6386090555393493 |
| ngcjob_appmetrics_loss | 2020-05-08T00:00:00Z | 1.356071743548495 |
| ngcjob_appmetrics_total_loss | 2020-05-04T06:00:00Z | 4.068851139706283 |
| ngcjob_appmetrics_total_loss | 2020-05-08T00:00:00Z | 1.9573122789348445 |
+-----------------------------------+----------------------+-----------------------+
How to get telemetry csv format data?¶
$ ngc base-command job telemetry --interval-time 90 --interval-unit HOUR --format_type csv 1120624
Time,App Metrics:job_rate,App Metrics:learn_rate,App Metrics:num_epochs,App Metrics:loss,App Metrics:total_loss,GPU_UTILIZATION,GPU_UTILIZATION_gpu_5,GPU_UTILIZATION_gpu_6,GPU_FI_PROF_PIPE_TENSOR_ACTIVE,GPU_FI_PROF_PIPE_TENSOR_ACTIVE_gpu_5,GPU_FI_PROF_PIPE_TENSOR_ACTIVE_gpu_6,GPU_FI_PROF_DRAM_ACTIVE,GPU_FI_PROF_DRAM_ACTIVE_gpu_5,GPU_FI_PROF_DRAM_ACTIVE_gpu_6,GPU_POWER_USAGE,GPU_POWER_USAGE_gpu_5,GPU_POWER_USAGE_gpu_6,GPU_FB_USED,GPU_FB_USED_gpu_5,GPU_FB_USED_gpu_6,GPU_FI_PROF_PCIE_RX_BYTES,GPU_FI_PROF_PCIE_RX_BYTES_gpu_5,GPU_FI_PROF_PCIE_RX_BYTES_gpu_6,GPU_FI_PROF_PCIE_TX_BYTES,GPU_FI_PROF_PCIE_TX_BYTES_gpu_5,GPU_FI_PROF_PCIE_TX_BYTES_gpu_6,GPU_NVLINK_BANDWIDTH_TOTAL,GPU_NVLINK_BANDWIDTH_TOTAL_gpu_5,GPU_NVLINK_BANDWIDTH_TOTAL_gpu_6,CPU_UTILIZATION,MEM_UTILIZATION
2020-05-04T06:00:00Z,351.479144517938,0.7806435847527863,38.453847984087396,2.6386090555393493,4.068851139706283,18.253698644530818,36.45377190221379,0.0,5.0732140045141945,10.155490175475455,0.0,8.480215505913197,16.963105877404956,0.0,70.64337745072271,95.24901714526119,46.06617461651623,6253.323733445449,12359.62638056169,156.0,5.681953140065174E8,1.1338959633084602E9,2256750.4371780045,7.030049576372905E7,1.3947316406226337E8,1222276.019745825,0.0,0.0,0.0,,1.0148062796099368E11
2020-05-08T00:00:00Z,373.13010860893604,0.018556020195032785,81.65085542400061,1.341255228094049,1.9361629716369715,19.79727223131477,39.70284463894965,0.0,5.537745098039208,11.087568157033791,0.0,9.228462377317365,18.477074235807915,0.0,72.7692570960701,99.37005337690641,46.052045951860336,6268.487725040916,12361.0,156.0,6.252954906637459E8,1.2455773122448087E9,2290148.461031833,7.701736719650654E7,1.5296292707322404E8,1237446.4907306435,0.0,0.0,0.0,0.8583701189035325,9.224660433413626E10
How to get result info?¶
$ ngc base-command result info 9999
How to remove result from job?¶
$ ngc base-command result remove 9999
How to download result?¶
$ ngc base-command result download 9999
How to create workspaces?¶
$ ngc base-command workspace create --ace nvidia-exmpl-1
Successfully created workspace with ID: 'e9UJrwAPTSmOQCxA'
Workspace Information
ID: e9UJrwAPTSmOQCxA
Name: None
Created By: ngccli
Size: 0 B
ACE: nvidia-exmpl-1
Org: robot
Description:
Shared with: None
How to list workspaces?¶
$ ngc base-command workspace list
+------------------+------------------+------------------+----------------+------------------+--------+--------------+-------+----------+
| Id | Name | Description | ACE | Creator Username | Shared | Created Date | Owned | Size |
+------------------+------------------+------------------+----------------+------------------+--------+--------------+-------+----------+
| e9UJrwAPTSmOQCxA | | | nvidia-exmpl-1 | Amy Smith | Yes | 2021-02-02 | No | 14.01 KB |
| | | | | | | 11:38:23 UTC | | |
| cizRgbYrQp7nFo | workspace-test | | nvidia-exmpl-1 | Bill Williams | Yes | 2021-02-19 | No | 0 B |
| | | | | | | 01:11:25 UTC | | |
| jHJx5rBoT3uTe-C8 | wkspc-example | | nvidia-exmpl-2 | Chad Brown | Yes | 2020-06-04 | No | 2.67 GB |
| | | | | | | 10:09:25 UTC | | |
+------------------+------------------+------------------+----------------+------------------+--------+--------------+-------+----------+
How to list workspaces using column arguments?¶
$ ngc base-command workspace list --column ace --column 'updated=Date'
+------------------------+----------------+-------------------------+
| Id | Ace | Date |
+------------------------+----------------+-------------------------+
| e9UJrwAPTSmOQCxA | nvidia-exmpl-1 | 2021-03-18 13:48:36 UTC |
| cizRgbYrQp7nFo | nvidia-exmpl-1 | 2021-02-25 12:33:09 UTC |
| jHJx5rBoT3uTe-C8 | nvidia-exmpl-2 | 2020-06-04 19:42:45 UTC |
+------------------------+----------------+-------------------------+
How to get workspace info?¶
$ ngc base-command workspace info workspace1
----------------------------------------------------
Workspace Information
ID: nwJF12QPQvKkPm5aIBTTkg
Name: workspace1
ACE: Staging
Org: nvidia
Description: workspace testing
Shared with: nvidia, nvidia/a10, nvidia/a11, nvidia/workspace_test
----------------------------------------------------