Make the MachineConfigPool wait survive real-world rollout times - #82
Open
Vincent056 wants to merge 1 commit into
Open
Make the MachineConfigPool wait survive real-world rollout times#82Vincent056 wants to merge 1 commit into
Vincent056 wants to merge 1 commit into
Conversation
The node-compliance e2e lane fails repeatedly with 'timed out waiting for MachineConfigPools ... master (Updated: 1/3, Unavailable: 1)' after a full remediation run. Three fixes: 1. Raise the MCP wait budget from 60 to 120 minutes. A remediation run reboots every node and the master pool always rolls serially (~10min per node on AWS), often in more than one render wave - the old budget sat exactly at the expected duration, so normal variance blew it. 2. Fix setPoolRollingPolicy: it compared pool.Spec.MaxUnavailable against the ADDRESS of a fresh local variable, which is never equal, so the intended MaxUnavailable=2 was never applied and worker reboots stayed fully serialized. Apply it by value comparison - and never to the master pool, where parallel reboots would threaten etcd quorum. 3. Diagnose instead of just dying: log per-pool progress transitions during the wait, and on timeout dump every node's machine-config daemon state annotations (state/reason/current/desired) so a drain hang is distinguishable from a reboot hang or a degraded daemon in the job output. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Collaborator
Author
|
/retest |
|
@Vincent056: The following tests failed, say
Full PR test history. Your PR dashboard. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. I understand the commands that are listed here. |
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
The
e2e-aws-openshift-node-compliancelane fails repeatedly with:after applying ~590 remediations cleanly (Applied=589, Error=0). Root cause is arithmetic, not flakiness: the master pool reboots its 3 nodes serially (~10 min each on AWS, often over more than one render wave) and the fixed 60-minute wait sat exactly at the expected duration.
Changes
setPoolRollingPolicy: it comparedpool.Spec.MaxUnavailableagainst the address of a fresh local variable — never equal — soMaxUnavailable: 2was silently never applied and worker reboots stayed serialized. Now compared by value, and explicitly skipped for the master pool (etcd quorum).state/reason/currentConfig/desiredConfig) are dumped, so a drain hang vs reboot hang vs degraded daemon is diagnosable from the job log instead of hand-waved as flake.Testing
go build ./...,go vet, gofmt clean.🤖 Generated with Claude Code