Retry on server load fail #4647
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
scheduler/cmd/agent/main.go
Outdated
| @@ -58,6 +58,10 @@ const ( | |||
| maxElapsedTimeReadySubServiceBeforeStart = 15 * time.Minute // 15 mins is the default MaxElapsedTime | |||
| // period for subservice ready "cron" | |||
| periodReadySubService = 60 * time.Second | |||
| // number of retries for loading a model onto a server | |||
| maxLoadRetryCount = 6 | |||
There was a problem hiding this comment.
nit: we could think about making it configurable in the future
RafalSkolasinski
pushed a commit
that referenced
this pull request
Feb 6, 2023
* fix type conversion and add test * Add a policy to retry with max count * wire up retry logic * change scope of helper functions * pass down defaults for retries * up number of retries on load * wire up correct unload const * remove unnecessary set of default value * bump maxLoadRetryCount to 10
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What this PR does / why we need it:
This PR adds the ability to retry loading / unloading models onto a server (control plane) with max number of attempts. This is different from the current usage of retries with backoff as we also cap retries with a number of attempts given that different models will have different latencies for loading / unloading.
We only do this on the control plane to get around the issue where we are loading an inference model and an associated explainer concurrently. So if an explainer tries to load first, it might fail if the underlying inference model has not loaded yet. With retries this should be able to eventually load the explainer.
Which issue(s) this PR fixes:
Fixes #4623
Special notes for your reviewer: