feat: writeback rollout updates to informer to prevent stale data #726
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
Codecov Report
@@ Coverage Diff @@
## master #726 +/- ##
==========================================
- Coverage 82.96% 82.91% -0.05%
==========================================
Files 95 95
Lines 7912 7926 +14
==========================================
+ Hits 6564 6572 +8
- Misses 947 951 +4
- Partials 401 403 +2
Continue to review full report at Codecov.
|
jessesuen
force-pushed
the
informer-writeback
branch
from
8e01130 to
c18ca32
Compare
jessesuen
added a commit
that referenced
this pull request
Sep 22, 2020
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Outdated suggestions cannot be applied.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.
Resolves #720
Currently the rollout controller may operate on stale information in the informer cache. This can easily happen when a rollout is requeued into the workqueue while already in the middle of reconciliation. When this happens, the controller immediately re-reconciles the rollout, but this time operating on a stale version of the rollout in the cache (it is stale because the changes to the rollout saved in first reconciliation, have not yet made the round trip into the cache).
At best, this simply duplicates work. At worst, it causes incorrect behavior (such as re-pausing after an unpause as in #720)
There is a technique which argo workflows uses that writes back the updated resource back into the informer cache. Using this technique, even if a rollout is immediately re-reconciled, it at least has the updates saved in the previous reconciliation.
Before this change, TestCanarySetCanaryScale could reproduce the double-pause problem in #720 quite easily. After this fix, I am unable to reproduce the problem.