You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Blake Eggleston (Jira)" <ji...@apache.org> on 2020/10/07 19:58:00 UTC
[jira] [Commented] (CASSANDRA-16113) Consolidate dead nodes check
in force repair
[ https://issues.apache.org/jira/browse/CASSANDRA-16113?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17209819#comment-17209819 ]
Blake Eggleston commented on CASSANDRA-16113:
---------------------------------------------
Just a few minor naming type stuff:
* RepairRunnable.java
** LiveNeighorsAndRanges
*** I think naming would make more sense if we changed the naming around for this a bit
**** Rename class back to `NeighorsAndRanges`, since it can contain dead neighbors if the force flag wasn't set.
**** Rename `allNeighbors` to `participants`, so we're not implying liveness or completeness
**** Rename `force` to `deadParticipantsExcluded` or something, to indicate the contents of `participants`
**** WDYT?
** `filterCommonRanges`
*** I'd reword the start of the new log message a bit. Something like "Skipping forced repair for ...", which is a bit clearer about what's happening and why
*** you only need `Sets.*difference*(commonRange.endpoints, endpoints)` when computing skipped replicas, since trans endpoints are a subset of all endpoints
* DistributedRepairUtils.java
** let's add a null check for `moreSuccessCritera` at the top of `assertParentRepairSuccess` so this fails quickly in the case of a null arg
> Consolidate dead nodes check in force repair
> --------------------------------------------
>
> Key: CASSANDRA-16113
> URL: https://issues.apache.org/jira/browse/CASSANDRA-16113
> Project: Cassandra
> Issue Type: Improvement
> Components: Local/Other
> Reporter: Yifan Cai
> Assignee: Yifan Cai
> Priority: Normal
> Fix For: 4.0-beta
>
> Time Spent: 4h 10m
> Remaining Estimate: 0h
>
> The check for dead nodes during force repair is duplicated in the normal and incremental repair. We could consolidate those 2 checks to make the code more dry.
> The check should throw a more meaningful error message to indicate that all neighbor nodes are down, instead of "java.lang.IllegalArgumentException: Endpoints can not be empty"
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org