kvserver: de-flake TestProtectedTimestamps #75436
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
|
This change is |
ajwerner
approved these changes
Jan 24, 2022
There was a problem hiding this comment.
Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status:complete! 1 of 0 LGTMs obtained (waiting on @arulajmani and @nvanbenschoten)
pkg/kv/kvserver/client_protectedts_test.go, line 97 at r1 (raw file):
count := 0 if err := conn.QueryRow( "SELECT COUNT(*) "+
linter says: SELECT count(*)
Fixed cockroachdb#75020. This test makes use of an "upsert-until-backpressure" primitive, where it continually writes blobs to a scratch table configured with a lower max range size (1<<18 from the default of 512<<20) until it experiences replica backpressure. Before cockroachdb#73876 (when using the system config span), the splitting off of the scratch table's ranges happened near instantly after the creation of the table itself. This changed slightly with the span configs infrastructure, where there's more of asynchrony built in and ranges may appear after a while longer. The test previously started upserting as soon as it created the table, being able to implicitly rely on the tight synchrony of already having the table's ranges carved out. This comes when deciding to record a replica's largest previous max range size -- something we only do if the replica's current size is larger than the new limit (see (*Replica).SetSpanCnfig). When we weren't upserting until the range applied the latest config, this was not possible. With span configs and its additional asynchrony, this assumption no longer held. It was possible then for the range to accept writes larger than the newest max range size, which unintentionally (for this test at least) triggers an escape hatch where we avoid backpressure when lowering a range's max size below its current size (cockroachdb#46863). The failure is easy to repro. This test runs in ~8s, and with the following we were reliably seeing timeouts where the test was waiting for backpressure to kick in (but it never did because of the escape hatch). dev test pkg/kv/kvserver \ -f TestProtectedTimestamps -v --show-logs \ --timeout 300s --ignore-cache --count 10 De-flaking this test is simple -- we just wait for the table's replicas to apply the latest config before issuing writes to it. Release note: None
irfansharif
force-pushed
the
220123.deflake-TestProtectedTimestamps
branch
from
49dd631
to
20161f9
Compare
irfansharif
commented
Jan 24, 2022
There was a problem hiding this comment.
Thanks!
bors r+
Reviewable status:
|
This PR was included in a batch that was canceled, it will be automatically retried |
|
Build failed (retrying...): |
|
Build failed (retrying...): |
|
Build failed (retrying...): |
|
Build failed (retrying...): |
|
Build succeeded: |
23 tasks
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.
Fixed #75020. This test makes use of an "upsert-until-backpressure"
primitive, where it continually writes blobs to a scratch table
configured with a lower max range size (1<<18 from the default of
512<<20) until it experiences replica backpressure. Before #73876 (when
using the system config span), the splitting off of the scratch table's
ranges happened near instantly after the creation of the table itself.
This changed slightly with the span configs infrastructure, where
there's more of asynchrony built in and ranges may appear after a while
longer.
The test previously started upserting as soon as it created the table,
being able to implicitly rely on the tight synchrony of already having
the table's ranges carved out. This comes when deciding to record a
replica's largest previous max range size -- something we only do if the
replica's current size is larger than the new limit (see
(*Replica).SetSpanCnfig). When we weren't upserting until the range
applied the latest config, this was not possible. With span configs and
its additional asynchrony, this assumption no longer held. It was
possible then for the range to accept writes larger than the newest max
range size, which unintentionally (for this test at least) triggers an
escape hatch where we avoid backpressure when lowering a range's max
size below its current size (#46863).
The failure is easy to repro. This test runs in ~8s, and with the
following we were reliably seeing timeouts where the test was waiting
for backpressure to kick in (but it never did because of the escape
hatch).
De-flaking this test is simple -- we just wait for the table's replicas
to apply the latest config before issuing writes to it.
Release note: None