Warning
The Autoscaler is currently in Beta and may experience changes, quirks, and downtime.
Both endptgroups and autogroups track a number of different metrics relating to their performance over time.
The list of the currently tracked metrics are:
nworkers: Current number of workers in the autogroup.totreqs: Total number of requests your autogroup has served.curload: The rate at which requested work units are coming in to the server (work units per second).capacity: The total number of requested work units your autogroup can handle.reliable: Average rate of success for requests sent to workers in your autogroup.reqrate: The rate at which new requests are coming into your autogroup.The /get_endpoint_stats/ and /get_autogroup_stats/ endpoints return these metrics in one of two modes.
id: ID of your endpoint groupendpoint: Name of your endpoint groupapi_key: The API_KEY corresponding to your endpoint group used for authentication purposesid: ID of your autogroupapi_key: The API_KEY corresponding to your autogroup used for authentication purposesIf you include only these parameters in your request, the endpoint will return the current value for all of these metrics.
1
2
3
4
5
6
7
8
9
10
{
"capacity": 8306.60345739313,
"curload": 9250.0,
"endpoint": "test-local",
"id": 28,
"nworkers": 44,
"reliable": 0.9942605208097545,
"reqrate": 31.918882596395715,
"totreqs": 8603534
}If you are interested in getting results for a specific metric over a specific time range, you can use this endpoint in "timeseries" mode by supplying these extra parameters:
metric_name: Name of the metric you are queryingt1: Start timepoint of the timeseriest2: End timepoint of the timeseries/get_autogroup_stats/ Endpoint in "Timeseries" Mode #1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
import requests
import json
stats_payload = {
"id": 28,
"metric_name": "nworkers",
"t1": 5000,
"t2": 5010,
"api_key": "API_KEY_HERE"
}
response = requests.post(
"https://run.vast.ai/get_autogroup_stats/",
headers={"Content-Type": "application/json"},
data=json.dumps(stats_payload),
timeout=4
)
if response.status_code != 200:
print(f"Failed to call /get_autogroup_stats/, response.status_code: {response.status_code}")
return/get_autogroup_stats/ Endpoint with cURL #1
curl https://run.vast.ai/get_autogroup_stats/ \n-X POST \n-d '{"id" : 28, "api_key" : "API_KEY_HERE", "metric_name" : "nworkers", "t1": 5000, "t2" : 5010}' \n-H 'Content-Type: application/json'1
2
3
4
5
6
7
8
9
{
"endpoint": "test-local",
"id": 28,
"nworkers": [
8.0,
8.0
],
"range_len": 2
}Note that the t1 and t2 timestamps are measured in seconds and are relative to the UNIX Epoch.