[GC] Fixed flaky GC stats end-to-end test #21919
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
agarwal-navin
requested review from
markfields,
a team,
pragya91,
jatgarg,
tyler-cai-microsoft,
kian-thompson and
rajatch-ff
and removed request for
a team
github-actions
bot
added
area: tests
agarwal-navin
changed the title
⯅ @fluid-example/bundle-size-tests: +245 Bytes
Baseline commit: 3f00c6d |
tyler-cai-microsoft
approved these changes
Jul 17, 2024
RishhiB
pushed a commit
to RishhiB/FluidFramework-1
that referenced
this pull request
Jul 18, 2024
## Bug and root-cause A GC test that validates deleted stats is flaky. The test uses sweep timeout of 200 ms. It does the following in sequence: 1. Runs GC wtih every thing referenced and validates that there unreferenced stats are 0. 2. Unreferences one data store and one blob, runs GC and validates they show as unreferenced. 3. Unreferences another data store and another blob, runs GC and validates they show as unreferenced. 4. Wait for sweep timeout, runs GC so that GC delete op is sent. 5. Run GC again and validate that the unreferenced data stores and blobs show up as deleted. If 200 ms have passed before step 3 is run, the unreferenced data store and blob will become sweep ready and will show up as deleted in GC stats. There was a change few months back that changed the logic to add sweep ready objects as deleted even when sweep is enabled (microsoft#19524) and the test probably became flaky then. ## Fix Broke the test above into 2: 1. The first one only validates unreferenced stats (steps 1 to 3 above). 2. The second one that only validates sweep ready and deleted stats. It unreferences objects, waits for sweep timeout so that nodes are sweep ready and validates that they show as deleted. It then runs GC again so that the GC sweep op has round tripped and these nodes are deleted. Since the test goes from unreferenced -> wait for sweep timeout -> validation, there is no intermediate GC run like bfore where it can unexpectedly become sweep ready. [AB#8350](https://dev.azure.com/fluidframework/235294da-091d-4c29-84fc-cdfc3d90890b/_workitems/edit/8350)
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.
Bug and root-cause
A GC test that validates deleted stats is flaky. The test uses sweep timeout of 200 ms. It does the following in sequence:
If 200 ms have passed before step 3 is run, the unreferenced data store and blob will become sweep ready and will show up as deleted in GC stats. There was a change few months back that changed the logic to add sweep ready objects as deleted even when sweep is enabled (#19524) and the test probably became flaky then.
Fix
Broke the test above into 2:
Since the test goes from unreferenced -> wait for sweep timeout -> validation, there is no intermediate GC run like bfore where it can unexpectedly become sweep ready.
AB#8350