[
https://issues.apache.org/jira/browse/SOLR-17231?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18039438#comment-18039438
]
Chris M. Hostetter commented on SOLR-17231:
-------------------------------------------
AB: I think your PR looks fine technically, but i don't know if people have any
strong philosophical opinions about moving more towards checks like
{{Replica.Type.isLeaderType}} instead of checks like
getUpdateHandler().getUpdateLog() != null
(The idea being that isLeaderType would tell you if there *should* be an
updateLog in this code path, and we explicitly fail if there is *NOT* an
updateLog at this point)
> SolrCore.bufferUpdatesIfConstructing fails on PULL replicas
> -----------------------------------------------------------
>
> Key: SOLR-17231
> URL: https://issues.apache.org/jira/browse/SOLR-17231
> Project: Solr
> Issue Type: Bug
> Reporter: Chris M. Hostetter
> Assignee: Andrzej Bialecki
> Priority: Major
> Labels: pull-request-available
> Time Spent: 10m
> Remaining Estimate: 0h
>
> If a "Slice" is in the {{CONSTRUCTION}} state, then SolrCore initialization
> for replicas of that Slice try to setup update buffering of the {{UpdateLog}}
> – but this makes no sense for {{PULL}} replicas which have no update logs,
> and causes a NPE....
> {code:java}
> private void bufferUpdatesIfConstructing(CoreDescriptor coreDescriptor) {
> ...
> if (slice.getState() == Slice.State.CONSTRUCTION) {
> // set update log to buffer before publishing the core
> getUpdateHandler().getUpdateLog().bufferUpdates();
> }
> }
> {code}
> I believe the fix is just for {{bufferUpdatesIfConstructing}} to check the
> replica type? (or do a null check on {{getUpdateLog()}} ?)
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
