You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We're observing constant flakiness on some of the agent's integration tests:
Tests know to fail:
TestInstallWithoutBasePath
Most likely the test setup needs /opt/Elastic/Agent to not exist, however instead the test ensuring it, the test fails if the folder is there. The folder might have been left there by another test that run before it. I thought I fixed it, I need to look for the PR/commit that should have fixed that.
When there is only one snapshot version for the current version the test fails as it relies on the fact there is at least 2 snapshots. It used to fail something like 1 is not greater than 1 (require.Greater(t, len(builds.Builds), 1))
Fixes and restructure of the test framework and tests
There are several fixes and changes to some tests and the test framework itself to make it easier to understand what is going on, to make it easier and more clear how to use the test framework and what to use and when. There all together on:
The "TestFleetManagedUpgrade" are not flaky, the tests have discovered an actual bug in the Elastic Agent. I have filed the issue that is causing the tests failures here - #3371
We're observing constant flakiness on some of the agent's integration tests:
Tests know to fail:
TestInstallWithoutBasePath
/opt/Elastic/Agent
to not exist, however instead the test ensuring it, the test fails if the folder is there. The folder might have been left there by another test that run before it. I thought I fixed it, I need to look for the PR/commit that should have fixed that.8.10:
TestStandaloneUpgradeToSpecificSnapshotBuild
require.Greater(t, len(builds.Builds), 1)
)Fixes and restructure of the test framework and tests
There are several fixes and changes to some tests and the test framework itself to make it easier to understand what is going on, to make it easier and more clear how to use the test framework and what to use and when. There all together on:
Test failing due to known issue:
TestFleetManagedUpgrade
any variation of/Upgrade_managed_agent_from_x.y.z_to_8.11.0-SNAPSHOT
:(fork/exec /opt/Elastic/Agent/elastic-agent: no such file or directory)
Click to expand
TestInstallWithEndpointSecurityAndRemoveEndpointIntegration/
unprotected
ORprtected
:This issue is reoccurring, opened a dedicated issue Flaky Test: TestInstallWithEndpointSecurityAndRemoveEndpointIntegration/* #3480
UPDATE 2: New occurrence after the fix for the actual bug has been merged ---> https://buildkite.com/elastic/elastic-agent/builds/3524#018ad0da-36cf-41bc-8161-7d1620878f38
This may be fixed by reintroducing the wait for the watcher in this PR Change the selection of downgrade version in TestUpgradeBrokenPackageVersion #3458
UPDATE 1: should be caused by the actual bug: Uninstall does not stop a running watcher after upgrade #3371
agent stuck in updating state on fleet
failed to collect diagnostics:
(fork/exec /opt/Elastic/Agent/elastic-agent: no such file or directory)
Click to expand
TestInstallAndUnenrollWithEndpointSecurity
:TestUpgradeBrokenPackageVersion
The text was updated successfully, but these errors were encountered: