Unnecessary extra reload on integration policy change under EA managed mode #41751
Labels
bug
impact:low
Long-term priority, unless it's a quick fix.
needs_team
Indicates that the issue/PR needs a Team:* label
APM Server version (
apm-server version
): 8.15.0+Description of the problem including expected versus actual behavior:
APM server reloader reloads runner (internal/beater/beater.go) 2 times instead of 1 when integration policy is changed.
This is likely due to nil check being earlier than the unchanged check in EA apm tracing PR: https://github.com/elastic/beats/pull/40030/files#diff-dd024b66f2ed585cbbfce1d8a0aeba55183ab0d04bed2a67f7738f62b5658e63R896-R908 . This causes apm tracing config nil -> nil to trigger a reload.
When this is fixed, we should also remove the workaround in https://github.com/elastic/apm-server/pull/14574/files#diff-e91b3485079439409a6365ec6aa74c2c009623463528bc9d541ca6494df8cf35R350
Steps to reproduce:
loaded input config
log line is accompanied by aloaded apm tracing config
even though apm tracing config has not changed. This implies 2 reloads are done instead of 1.Provide logs (if relevant):
apm-server log lines from new to old. Everything is triggered by
BeatV2Manager.unitListen
, followed byloaded apm tracing config
, thenloaded input config
.The text was updated successfully, but these errors were encountered: