[
https://issues.apache.org/jira/browse/SOLR-10169?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16436255#comment-16436255
]
Erick Erickson commented on SOLR-10169:
---------------------------------------
PeerSync.alreadyInSync() eventually calls HttpShardHandler.take(), which has
this line:
if (bailOnError && rsp.getException() != null) return rsp;
Then takes the response and throws an NPE on line 391:
Object replicaFingerprint =
srsp.getSolrResponse().getResponse().get("fingerprint");
It's not clear to me whether it's the getSolrResponse() or
getSolrResponse().getResponse() that throws the error.
The problem here is that the exception bypasses the rest of the peer sync
logic and goes into full sync.
I'll attach a patch shortly that tests both for null. I Haven't done anything
except compile it yet. It may still go into full sync, but at least there'll be
a chance to recover.
I'll commit this over the weekend unless there are objections.
The line numbers match up reasonably between the various versions of that file
for the dates of the JIRAs.
> PeerSync will hit an NPE on no response errors when looking for fingerprint.
> ----------------------------------------------------------------------------
>
> Key: SOLR-10169
> URL: https://issues.apache.org/jira/browse/SOLR-10169
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Affects Versions: 6.6, 7.0
> Reporter: Mark Miller
> Assignee: Erick Erickson
> Priority: Major
> Attachments: SOLR-10169.patch
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
