[
https://issues.apache.org/jira/browse/ZOOKEEPER-5080?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Ivan Khanas updated ZOOKEEPER-5080:
-----------------------------------
Component/s: tests
Affects Version/s: 3.10.0
Description:
Seen on CI for PR #2435, one failure out of 3262:
{noformat}
SnapshotAndRestoreCommandTest.testSnapshotAndRestoreCommand_streaming:168
-> validateSnapshotMetrics:398 expected: <true> but was: <false>
{noformat}
The assertion:
{noformat}
assertTrue((Double) metrics.get("avg_snapshottime") > 0.0);
{noformat}
avg_snapshottime is fed by ZooKeeperServer.takeSnapshot:
{noformat}
long elapsed = Time.currentElapsedTime() - start;
ServerMetrics.getMetrics().SNAPSHOT_TIME.add(elapsed);
{noformat}
elapsed counts whole milliseconds. The database in this test holds a handful of
znodes, so on a fast runner the snapshot finishes inside one millisecond,
elapsed is 0, and the average never gets above zero. Nothing about the snapshot
went wrong, the clock just was not fine grained enough to notice it happened.
validateRestoreMetrics one method down has the same problem waiting:
avg_restore_time > 0.0, fed by RESTORE_TIME.add(elapsed) at ZooKeeperServer:644.
I have not seen it fail yet, but it will eventually, for the same reason.
The same CI job also failed on master at 4ee3b27be and passed again on the next
commit, so this is not something a branch introduced.
Priority: Minor (was: Major)
> Flaky test:
> SnapshotAndRestoreCommandTest.testSnapshotAndRestoreCommand_streaming
> ---------------------------------------------------------------------------------
>
> Key: ZOOKEEPER-5080
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-5080
> Project: ZooKeeper
> Issue Type: Bug
> Components: tests
> Affects Versions: 3.10.0
> Reporter: Ivan Khanas
> Priority: Minor
>
> Seen on CI for PR #2435, one failure out of 3262:
> {noformat}
> SnapshotAndRestoreCommandTest.testSnapshotAndRestoreCommand_streaming:168
> -> validateSnapshotMetrics:398 expected: <true> but was: <false>
> {noformat}
> The assertion:
> {noformat}
> assertTrue((Double) metrics.get("avg_snapshottime") > 0.0);
> {noformat}
> avg_snapshottime is fed by ZooKeeperServer.takeSnapshot:
> {noformat}
> long elapsed = Time.currentElapsedTime() - start;
> ServerMetrics.getMetrics().SNAPSHOT_TIME.add(elapsed);
> {noformat}
> elapsed counts whole milliseconds. The database in this test holds a handful
> of
> znodes, so on a fast runner the snapshot finishes inside one millisecond,
> elapsed is 0, and the average never gets above zero. Nothing about the
> snapshot
> went wrong, the clock just was not fine grained enough to notice it happened.
> validateRestoreMetrics one method down has the same problem waiting:
> avg_restore_time > 0.0, fed by RESTORE_TIME.add(elapsed) at
> ZooKeeperServer:644.
> I have not seen it fail yet, but it will eventually, for the same reason.
> The same CI job also failed on master at 4ee3b27be and passed again on the
> next
> commit, so this is not something a branch introduced.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
