Fixed HA Tracker jitter causing unnecessary CAS operations #1861
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
pstibrany
approved these changes
Nov 28, 2019
pkg/distributor/ha_tracker_test.go
Outdated
| if r.GetReplica() != replica { | ||
| err = fmt.Errorf("replicas did not match: %s != %s", r.GetReplica(), replica) | ||
| continue outer | ||
| } | ||
| if timestamp.Time(r.GetReceivedAt()).Equal(expected) { | ||
| } else if !timestamp.Time(r.GetReceivedAt()).Equal(expected) { |
There was a problem hiding this comment.
Tiny detail, but using 'else if' when previous 'if' body does continue, break or return is not very nice. Previous version was clearer.
pstibrany
reviewed
Nov 28, 2019
| if r.GetReplica() != replica { | ||
| err = fmt.Errorf("replicas did not match: %s != %s", r.GetReplica(), replica) | ||
| continue outer | ||
| } | ||
| if timestamp.Time(r.GetReceivedAt()).Equal(expected) { | ||
| if !timestamp.Time(r.GetReceivedAt()).Equal(expected) { |
pstibrany
approved these changes
Nov 28, 2019
cstyan
reviewed
Nov 29, 2019
| }) | ||
| assert.NoError(t, err) | ||
|
|
||
| // Write the first time. | ||
| mtime.NowForce(start) |
There was a problem hiding this comment.
was there something that necessitated this change?
There was a problem hiding this comment.
It avoids flaky tests.
| }) | ||
| assert.NoError(t, err) | ||
|
|
||
| // Write the first time. | ||
| mtime.NowForce(startTime) |
There was a problem hiding this comment.
It avoids flaky tests.
bboreham
approved these changes
Nov 29, 2019
|
@gouthamve Do we want this in the |
Signed-off-by: Marco Pracucci <marco@pracucci.com>
Signed-off-by: Marco Pracucci <marco@pracucci.com>
Signed-off-by: Marco Pracucci <marco@pracucci.com>
Signed-off-by: Marco Pracucci <marco@pracucci.com>
pracucci
force-pushed
the
fix-ha-tracker-cas-contention
branch
from
47e140c to
9e37d00
Compare
gouthamve
approved these changes
Dec 2, 2019
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.
What this PR does:
Today me and @pstibrany spent few hours debugging an expected pattern on the number of CAS operations done by the distributors when the HA tracker is enabled.
We found out that the PR #1748 - which introduced the update timeout jitter - also introduced a time window (long as much as the jitter) during which every request does a CAS but the CAS operation itself doesn't update the replica updated timestamp, because the CAS is triggered if
now - receivedAt >= updateTimeoutbut then the CAS function is a noop ifnow - receivedAt < updateTimeout + jitter.While adding tests, I've also realized that the
checkReplicaTimestamp()was broken (tests were actually failing) so I've fixed it.Which issue(s) this PR fixes:
No issue
Checklist
CHANGELOG.mdupdated - the order of entries should be[CHANGE],[FEATURE],[ENHANCEMENT],[BUGFIX]