ILM: Moving full inspection of all indices to MasterFailOver condition #87616
Conversation
elasticsearchmachine
added
v8.4.0
external-contributor
gwbrown
added
the
:Data Management/ILM+SLM
elasticmachine
added
the
Team:Data Management
|
Pinging @elastic/es-data-management (Team:Data Management) |
| @@ -208,9 +212,9 @@ void onMaster(ClusterState clusterState) { | |||
| } catch (Exception e) { | |||
| if (logger.isTraceEnabled()) { | |||
| logger.warn( | |||
| () -> format( | |||
| new ParameterizedMessage( | |||
There was a problem hiding this comment.
Thise lambda to ParameterizedMessage looks like a bad merge to me, please undo it.
| @@ -220,9 +224,9 @@ void onMaster(ClusterState clusterState) { | |||
| ); | |||
| } else { | |||
| logger.warn( | |||
| () -> format( | |||
| new ParameterizedMessage( | |||
| @@ -243,6 +247,7 @@ void onMaster(ClusterState clusterState) { | |||
| } | |||
| } | |||
|
|
|||
|
|
|||
There was a problem hiding this comment.
Please remove errant whitespace change.
I see. I will try to fix the details you mentioned and make sure it does compile. I'm sorry, I believed it had passed gradle check, but I was mistaken. |
|
No worries, best of luck. |
The previous version of the PR did not compile due to a variable that was being called on MasterFailOver, but was not set. I fixed this problem, plus the mentioned errant whitespace and bad merge.
|
Hi @PDTCCLF, thank you for working on this. I see there hasn't been any progress in this PR since July, I hope you do not mind if I give it a try myself. Let me know if you are still actively working on it. Then I will look for something else. |
Not at all. You can work on it. Thanks for asking. |
This is my attempt at solving issue #80407. As per the description of the issue, the costly loop that inspects all indices of a cluster state is no longer being executed by default. A full inspection only happens when an exception occurs to the master, inspired by SnapshotService.