vast-ai/vast-cli / vastai
vastai
FlaggedVast.ai CLI to manage GPU instances, volumes, serverless endpoints, and billing.
cli/commands/endpoints.py
View on GitHub"""CLI commands for managing endpoints and workergroups."""
import json
import argparse
from vastai.cli.parser import argument
from vastai.cli.display import deindent
from vastai.api import endpoints as endpoints_api
from vastai.cli.utils import get_parser as _get_parser, get_client # noqa: F401
parser = _get_parser()
# ---------------------------------------------------------------------------
# endpoints
# ---------------------------------------------------------------------------
@parser.command(
argument("--min_load", help="minimum floor load in perf units/s (token/s for LLms)", type=float, default=0.0),
argument("--min_cold_load", help="minimum floor load in perf units/s (token/s for LLms), but allow handling with cold workers", type=float, default=0.0),
argument("--target_util", help="target capacity utilization (fraction, max 1.0, default 0.9)", type=float, default=0.9),
argument("--cold_mult", help="cold/stopped instance capacity target as multiple of hot capacity target (default 2.5)", type=float, default=2.5),
argument("--cold_workers", help="min number of workers to keep 'cold' when you have no load (default 5)", type=int, default=5),
argument("--max_workers", help="max number of workers your endpoint group can have (default 20)", type=int, default=20),
argument("--endpoint_name", help="deployment endpoint name (allows multiple autoscale groups to share same deployment endpoint)", type=str),
argument("--max_queue_time", help="maximum seconds requests may be queued on each worker (default 30.0)", type=float),
argument("--target_queue_time", help="target seconds for the queue to be cleared (default 10.0)", type=float),
argument("--inactivity_timeout", help="seconds of no traffic before the endpoint can scale to zero active workers", type=int),
argument("--auto_instance", help=argparse.SUPPRESS, type=str, default="prod"),
usage="vastai create endpoint [OPTIONS]",
help="Create a new endpoint group",
epilog=deindent("""
Create a new endpoint group to manage many autoscaling groups
Example: vastai create endpoint --target_util 0.9 --cold_mult 2.0 --endpoint_name "LLama"
"""),
)
def create__endpoint(args):
"""Create a new endpoint group."""
if args.explain:
print("request json: ")
print({
"client_id": "me", "min_load": args.min_load, "min_cold_load": args.min_cold_load,
"target_util": args.target_util, "cold_mult": args.cold_mult,
"cold_workers": args.cold_workers, "max_workers": args.max_workers,
"endpoint_name": args.endpoint_name, "autoscaler_instance": args.auto_instance,
})
client = get_client(args)
result = endpoints_api.create_endpoint(
client, min_load=args.min_load, min_cold_load=args.min_cold_load,
target_util=args.target_util, cold_mult=args.cold_mult,
cold_workers=args.cold_workers, max_workers=args.max_workers,
endpoint_name=args.endpoint_name, auto_instance=args.auto_instance,
max_queue_time=args.max_queue_time, target_queue_time=args.target_queue_time,
inactivity_timeout=args.inactivity_timeout,
)
if args.raw:
return result
print("create endpoint {}".format(result))
@parser.command(
usage="vastai show endpoints [--api-key API_KEY]",
help="Display user's current endpoint groups",
epilog=deindent("""
Example: vastai show endpoints
"""),
)
def show__endpoints(args):
"""Display user's current endpoint groups."""
if args.explain:
print("request json: ")
print({"client_id": "me", "api_key": args.api_key})
client = get_client(args)
result = endpoints_api.show_endpoints(client)
if isinstance(result, dict) and "error" in result:
print(result["error"])
return
if args.raw:
return result
else:
print(json.dumps(result, indent=1, sort_keys=True))
@parser.command(
argument("id", help="id of endpoint group to update", type=int),
argument("--min_load", help="minimum floor load in perf units/s (token/s for LLms)", type=float),
argument("--min_cold_load", help="minimum floor load in perf units/s (token/s for LLms), but allow handling with cold workers", type=float),
argument("--endpoint_state", help="active, suspended, or stopped", type=str),
argument("--auto_instance", help=argparse.SUPPRESS, type=str, default="prod"),
argument("--target_util", help="target capacity utilization (fraction, max 1.0, default 0.9)", type=float),
argument("--cold_mult", help="cold/stopped instance capacity target as multiple of hot capacity target (default 2.5)", type=float),
argument("--cold_workers", help="min number of workers to keep 'cold' when you have no load (default 5)", type=int),
argument("--max_workers", help="max number of workers your endpoint group can have (default 20)", type=int),
argument("--endpoint_name", help="deployment endpoint name (allows multiple workergroups to share same deployment endpoint)", type=str),
argument("--max_queue_time", help="maximum seconds requests may be queued on each worker (default 30.0)", type=float),
argument("--target_queue_time", help="target seconds for the queue to be cleared (default 10.0)", type=float),
argument("--inactivity_timeout", help="seconds of no traffic before the endpoint can scale to zero active workers", type=int),
usage="vastai update endpoint ID [OPTIONS]",
help="Update an existing endpoint group",
epilog=deindent("""
Example: vastai update endpoint 4242 --min_load 100 --target_util 0.9 --cold_mult 2.0 --endpoint_name "LLama"
"""),
)
def update__endpoint(args):
"""Update an existing endpoint group."""
client = get_client(args)
result = endpoints_api.update_endpoint(
client, id=args.id,
min_load=args.min_load, min_cold_load=args.min_cold_load,
target_util=args.target_util, cold_mult=args.cold_mult,
cold_workers=args.cold_workers, max_workers=args.max_workers,
endpoint_name=args.endpoint_name, endpoint_state=args.endpoint_state,
auto_instance=args.auto_instance,
max_queue_time=args.max_queue_time, target_queue_time=args.target_queue_time,
inactivity_timeout=args.inactivity_timeout,
)
if args.raw:
return result
print("update endpoint {}".format(result))
@parser.command(
argument("id", help="id of endpoint group to delete", type=int),
usage="vastai delete endpoint ID ",
help="Delete an endpoint group",
epilog=deindent("""
Example: vastai delete endpoint 4242
"""),
)
def delete__endpoint(args):
"""Delete an endpoint group."""
id = args.id
if args.explain:
print("request json: ")
print({"client_id": "me", "endptjob_id": args.id})
client = get_client(args)
result = endpoints_api.delete_endpoint(client, id=id)
print("delete endpoint {}".format(result))
@parser.command(
argument("id", help="id of endpoint group to fetch logs from", type=int),
argument("--level", help="log detail level (0 to 3)", type=int, default=1),
argument("--tail", help="", type=int, default=None),
usage="vastai get endpt-logs ID [--api-key API_KEY]",
help="Fetch logs for a specific serverless endpoint group",
epilog=deindent("""
Example: vastai get endpt-logs 382
"""),
)
def get__endpt_logs(args):
"""Fetch logs for a specific serverless endpoint group."""
if args.explain:
print(f"Fetching endpoint logs for id={args.id}")
client = get_client(args)
rj = endpoints_api.get_endpt_logs(client, id=args.id, level=args.level, tail=args.tail)
levels = {0: "info0", 1: "info1", 2: "trace", 3: "debug"}
if isinstance(rj, dict) and "error" in rj:
print(rj["error"])
return
if args.raw:
return rj
else:
dbg_lvl = levels[args.level]
if rj and dbg_lvl:
print(rj[dbg_lvl])
@parser.command(
argument("id", help="id of endpoint group whose workers to list", type=int),
usage="vastai get endpt-workers ID [--api-key API_KEY]",
help="List workers on an endpoint group with status and measured_perf",
epilog=deindent("""
Example: vastai get endpt-workers 21979
"""),
)
def get__endpt_workers(args):
"""List workers on a specific serverless endpoint group."""
client = get_client(args)
rj = endpoints_api.get_endpoint_workers(client, id=args.id)
if isinstance(rj, dict) and "error" in rj:
print(rj["error"])
return
workers = rj if isinstance(rj, list) else rj.get("workers", [])
if args.raw:
return workers
if not workers:
print("No workers found.")
return
for w in workers:
print(f"id={w.get('id')} status={w.get('status')} "
f"gpu_name={w.get('gpu_name')} measured_perf={w.get('measured_perf')}")
# ---------------------------------------------------------------------------
# workergroups
# ---------------------------------------------------------------------------
@parser.command(
argument("--template_hash", help="template hash (required, but **Note**: if you use this field, you can skip search_params, as they are automatically inferred from the template)", type=str),
argument("--template_id", help="template id (optional)", type=int),
argument("-n", "--no-default", action="store_true", help="Disable default search param query args"),
argument("--launch_args", help="launch args string for create instance ex: \"--onstart onstart_wget.sh --env '-e ONSTART_PATH=https://s3.amazonaws.com/vast.ai/onstart_OOBA.sh' --image atinoda/text-generation-webui:default-nightly --disk 64\"", type=str),
argument("--endpoint_name", help="deployment endpoint name (allows multiple workergroups to share same deployment endpoint)", type=str),
argument("--endpoint_id", help="deployment endpoint id (allows multiple workergroups to share same deployment endpoint)", type=int),
argument("--gpu_ram", help="estimated GPU RAM req (independent of search string)", type=float),
argument("--search_params", help="search param string for search offers ex: \"gpu_ram>=23 num_gpus=2 gpu_name=RTX_4090 inet_down>200 direct_port_count>2 disk_space>=64\"", type=str),
argument("--auto_instance", help=argparse.SUPPRESS, type=str, default="prod"),
usage="vastai create workergroup [OPTIONS]",
help="Create a new autoscale group",
epilog=deindent("""
Create a new autoscaling group to manage a pool of worker instances.
Example: vastai create workergroup --template_hash HASH --endpoint_name "LLama"
"""),
)
def create__workergroup(args):
"""Create a new workergroup."""
if args.explain:
print("request json: ")
print({"template_hash": args.template_hash, "search_params": args.search_params})
client = get_client(args)
try:
result = endpoints_api.create_workergroup(
client, template_hash=args.template_hash, template_id=args.template_id,
no_default=args.no_default, launch_args=args.launch_args,
endpoint_name=args.endpoint_name, endpoint_id=args.endpoint_id,
gpu_ram=args.gpu_ram, search_params=args.search_params,
auto_instance=args.auto_instance,
)
print("workergroup create {}".format(result))
except Exception as e:
print(f"Error creating workergroup: {e}")
@parser.command(
usage="vastai show workergroups [--api-key API_KEY]",
help="Display user's current workergroups",
epilog=deindent("""
Example: vastai show workergroups
"""),
)
def show__workergroups(args):
"""Display user's current workergroups."""
if args.explain:
print("request json: ")
print({"client_id": "me", "api_key": args.api_key})
client = get_client(args)
result = endpoints_api.show_workergroups(client)
if isinstance(result, dict) and "error" in result:
print(result["error"])
return
if args.raw:
return result
else:
print(json.dumps(result, indent=1, sort_keys=True))
@parser.command(
argument("id", help="id of autoscale group to update", type=int),
argument("--gpu_ram", help="estimated GPU RAM req (independent of search string)", type=float),
argument("--template_hash", help="template hash (**Note**: if you use this field, you can skip search_params, as they are automatically inferred from the template)", type=str),
argument("--template_id", help="template id", type=int),
argument("--search_params", help="search param string for search offers ex: \"gpu_ram>=23 num_gpus=2 gpu_name=RTX_4090 inet_down>200 direct_port_count>2 disk_space>=64\"", type=str),
argument("-n", "--no-default", action="store_true", help="Disable default search param query args"),
argument("--launch_args", help="launch args string for create instance ex: \"--onstart onstart_wget.sh --env '-e ONSTART_PATH=https://s3.amazonaws.com/public.vast.ai/onstart_OOBA.sh' --image atinoda/text-generation-webui:default-nightly --disk 64\"", type=str),
argument("--endpoint_name", help="deployment endpoint name (allows multiple workergroups to share same deployment endpoint)", type=str),
argument("--endpoint_id", help="deployment endpoint id (allows multiple workergroups to share same deployment endpoint)", type=int),
usage="vastai update workergroup WORKERGROUP_ID --endpoint_id ENDPOINT_ID [options]",
help="Update an existing autoscale group",
epilog=deindent("""
Example: vastai update workergroup 4242 --search_params \"gpu_ram>=23 num_gpus=2 gpu_name=RTX_4090 inet_down>200 direct_port_count>2 disk_space>=64\" --launch_args \"--onstart onstart_wget.sh --env '-e ONSTART_PATH=https://s3.amazonaws.com/public.vast.ai/onstart_OOBA.sh' --image atinoda/text-generation-webui:default-nightly --disk 64\" --gpu_ram 32.0 --endpoint_name "LLama" --endpoint_id 2
"""),
)
def update__workergroup(args):
"""Update an existing workergroup."""
client = get_client(args)
result = endpoints_api.update_workergroup(
client, id=args.id,
gpu_ram=args.gpu_ram,
template_hash=args.template_hash, template_id=args.template_id,
search_params=args.search_params, no_default=args.no_default,
launch_args=args.launch_args, endpoint_name=args.endpoint_name,
endpoint_id=args.endpoint_id,
)
if args.raw:
return result
print("workergroup update {}".format(result))
@parser.command(
argument("id", help="id of workergroup to update workers for", type=int),
argument("--cancel", action="store_true", help="cancel an in-progress update for the workergroup"),
usage="vastai update workers WORKERGROUP_ID [--cancel]",
help="Trigger a rolling update of all workers in a workergroup, or cancel an in-progress update",
epilog=deindent("""
Starts a rolling update of all workers in the specified workergroup. The autoscaler
will cycle through workers, updating them while maintaining capacity.
Use --cancel to cancel an update that is currently in progress.
Examples:
vastai update workers 4242
vastai update workers 4242 --cancel
"""),
)
def update__workers(args):
"""Trigger a rolling update of workers in a workergroup."""
client = get_client(args)
result = endpoints_api.update_workers(client, id=args.id, cancel=args.cancel)
if args.raw:
return result
if result.get("success"):
if result.get("cancelled"):
print(f"Update cancelled for workergroup {args.id}")
else:
print(f"Update started for workergroup {args.id} ({result.get('workers_to_update', 0)} workers)")
else:
print(f"Error: {result.get('error_msg', 'unknown error')}")
@parser.command(
argument("id", help="id of group to delete", type=int),
usage="vastai delete workergroup ID ",
help="Delete a workergroup group",
epilog=deindent("""
Note that deleting a workergroup doesn't automatically destroy all the instances that are associated with your workergroup.
Example: vastai delete workergroup 4242
"""),
)
def delete__workergroup(args):
"""Delete a workergroup."""
id = args.id
if args.explain:
print("request json: ")
print({"client_id": "me", "autojob_id": args.id})
client = get_client(args)
result = endpoints_api.delete_workergroup(client, id=id)
print("workergroup delete {}".format(result))
@parser.command(
argument("id", help="id of workergroup to fetch logs from", type=int),
argument("--level", help="log detail level (0 to 3)", type=int, default=1),
argument("--tail", help="", type=int, default=None),
usage="vastai get wrkgrp-logs ID [--api-key API_KEY]",
help="Fetch logs for a specific serverless worker group",
epilog=deindent("""
Example: vastai get endpt-logs 382
"""),
)
def get__wrkgrp_logs(args):
"""Fetch logs for a specific serverless worker group."""
if args.explain:
print(f"Fetching workergroup logs for id={args.id}")
client = get_client(args)
rj = endpoints_api.get_wrkgrp_logs(client, id=args.id, level=args.level, tail=args.tail)
levels = {0: "info0", 1: "info1", 2: "trace", 3: "debug"}
if isinstance(rj, dict) and "error" in rj:
print(rj["error"])
return
if args.raw:
return rj
else:
dbg_lvl = levels[args.level]
if rj and dbg_lvl:
print(rj[dbg_lvl])
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |