Do not memoize system-probe sock path error #32728
Conversation
The system-probe socket may not exist temporarily due to expected races during startup. This error should not be memoized. This restores the behaviour to how it was before #31150.
vitkyrka
added
changelog/no-changelog
team/usm
github-actions
bot
added
the
short review
| sysProbeUtil := newSystemProbe(path) | ||
| err = sysProbeUtil.initRetry.SetupRetrier(&retry.Config{ //nolint:errcheck | ||
| err := sysProbeUtil.initRetry.SetupRetrier(&retry.Config{ //nolint:errcheck |
There was a problem hiding this comment.
if we get an error (a temporary error) here, we will still memoize it
I don't think the memoize is a good idea here
There was a problem hiding this comment.
The error returning from this function is just for configuration issues of the retrier https://github.com/DataDog/datadog-agent/blob/7.61.x/pkg/util/retry/retrier.go#L30
There was a problem hiding this comment.
And before the PR this was done in a sync.Once and the error was ignored: https://github.com/DataDog/datadog-agent/blob/7.60.1/pkg/process/net/common.go#L78-L88
github-actions
bot
added
medium review
Test changes on VMUse this command from test-infra-definitions to manually test this PR changes on a VM: inv create-vm --pipeline-id=52329758 --os-family=ubuntuNote: This applies to commit 77f1d59 |
|
/merge |
Devflow running:
|
|
/trigger-ci --variable RUN_ALL_BUILDS=true --variable RUN_KITCHEN_TESTS=true --variable RUN_E2E_TESTS=on --variable RUN_UNIT_TESTS=on --variable RUN_KMT_TESTS=on |
Devflow running:
|
What does this PR do?
Do not memoize system-probe sock path error
The system-probe socket may not exist temporarily due to expected races
during startup. This error should not be memoized.
This restores the behaviour to how it was before #31150.
Motivation
Incident
Describe how you validated your changes
Tested manually by starting the agent first and then system-probe later. The service discovery check initially
fails but then succeeds on later runs after system-probe has been started. Without this PR it always fails even after system-probe starts.
Possible Drawbacks / Trade-offs
Additional Notes