chore(gcp): bump monolithic GCP deployment to Platforma chart 4.2.2#79
Open
mike-ainsel wants to merge 3 commits into
Open
chore(gcp): bump monolithic GCP deployment to Platforma chart 4.2.2#79mike-ainsel wants to merge 3 commits into
mike-ainsel wants to merge 3 commits into
Conversation
Vendor the v4.2.0 chart (from the release tag, so it matches the 4.2.0 backend image) in place of 4.1.5. Verified: helm lint passes and the chart renders against the monolith's values (gcs main storage, Filestore workspace, dedicated Kueue with the batch ComputeClass nodeSelector, GPU disabled) — batch flavor keeps cloud.google.com/compute-class=platforma-batch, --runner-max-ram-request stays 484Gi, and the job template's ResourceID field is satisfied by the paired 4.2.0 backend. Update the upgrade runbook to 4.2.0 and note that the one-time re-login only applies when coming from 3.5.0 (4.1.x already has the master secret).
Re-vendor the v4.2.2 chart (from the release tag) over 4.2.0. GPU stays disabled (pools.gpu.enabled=false) — the monolith has no GPU node pools, so the chart's GPU plumbing remains dormant (--runner-gpu-available=disabled). Infrastructure/terraform unchanged. Verified: helm lint passes and the chart renders against the monolith's values (gcs main, Filestore workspace, dedicated Kueue with the batch ComputeClass nodeSelector, GPU off) — batch keeps cloud.google.com/compute-class=platforma-batch and the 484Gi cap. Runbook renamed to UPGRADE-monolith-to-4.2.2.md.
Chart 4.2.2 pairs with the 4.2.2 backend image (chart default handles it unless platforma_image_override is pinned). Also document that OOM job retries are off by default in 4.2.x, so out-of-memory jobs fail fast with their original error instead of clogging the batch queue.
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.
What
Bumps the vendored chart for the monolithic GCP deployment to 4.2.2 (from the
v4.2.2release tag, matching the released 4.2.2 backend image). Supersedes #78 (which targeted 4.2.0). Base ischore/gcp-monolith-backport, so the diff is the full chart bump + runbook.Scope (deliberately narrow)
pools.gpu.enabled=false; the 4.2.x chart's GPU plumbing stays dormant (--runner-gpu-available=disabled). Enabling functional GPU would require adding GPU node pools + quota, which is out of scope here.isOOMFailurestill marker-based atv4.2.2); that remains a later release.Verification
helm lint— passes.helm templateagainst the monolith's values (gcs main storage, Filestore workspace, dedicated Kueue with the batch ComputeClass nodeSelector, GPU off) renders 27 objects with:cloud.google.com/compute-class: platforma-batch--runner-max-ram-request=484Gipreserved--runner-gpu-available=disabled<< .ResourceID >>present (satisfied by the paired 4.2.2 backend)Notes
UPGRADE-monolith-to-4.2.2.md; re-login note applies only when upgrading from 3.5.0.