Stop idle non-latest sampler pools even when replicas linger - #6
Open
cnnradams wants to merge 1 commit into
Open
Stop idle non-latest sampler pools even when replicas linger#6cnnradams wants to merge 1 commit into
cnnradams wants to merge 1 commit into
Conversation
A base-model pool used by one eval kept an H200 replica alive for 12 hours after its last request: the Flash health probes keep the web server warm, and the cleaner skipped any idle pool that still reported replicas. Keep the 5 minute grace for pools whose replicas may be finishing a generation, but stop the pool once it has been idle for 30 minutes regardless. Co-authored-by: Cursor <cursoragent@cursor.com>
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
_cleanup_fft_poolsskipped every idle non-latest pool that still had replicas. A replica never scales to zero on its own here: the Flash pool's health probes (GET /metrics,GET /server_infoevery few seconds) count as web-server traffic, so the container stays warm pastscaledown_windowforever. Observed:lilo-fft-…-v0(baseqwen3_5_9b_full_64kpool used by a TB-Lite eval) held one H200 for 12 h after the last sample;fft_pool_touchwas 06:46 UTC, container still serving/metricsat 14:57 UTC.FFT_POOL_ORPHAN_TIMEOUT) whether or not replicas are reported.Test plan
tests/providers/test_modal_app.py::test_cleanup_stops_long_idle_pool_with_lingering_replicasMade with Cursor