[
https://issues.apache.org/jira/browse/SOLR-9555?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15828949#comment-15828949
]
Cao Manh Dat edited comment on SOLR-9555 at 1/19/17 12:37 AM:
--------------------------------------------------------------
[~romseygeek]: In basically the reason for a leader put a replica into LIR
state is "it failed to send an update to the replica". FYI: "leader won't send
an update to the replica if the state of replica is down". So there are two
possible case here
- before sending update, the state of replica is active, then the replica go
down right after, so the update will be failed to send to the replica.
- before sending update, the state of replica is recovery, but the underlying
socket to replica is closed because the replica have just restarted, so the
update will be failed to send to the replica.
In the second case, although the replica already in recovery state, putting the
replica into LIR is a right thing to do. Because the buffering log doesn't
contain the failed update.
was (Author: caomanhdat):
@Alan : In basically the reason for a leader put a replica into LIR state is
"it failed to send an update to the replica". FYI: "leader won't send an update
to the replica if the state of replica is down". So there are two possible case
here
- before sending update, the state of replica is active, then the replica go
down right after, so the update will be failed to send to the replica.
- before sending update, the state of replica is recovery, but the underlying
socket to replica is closed because the replica have just restarted, so the
update will be failed to send to the replica.
In the second case, although the replica already in recovery state, putting the
replica into LIR is a right thing to do. Because the buffering log doesn't
contain the failed update.
> Recovery can hang if a node is put into LIR as it is starting up
> ----------------------------------------------------------------
>
> Key: SOLR-9555
> URL: https://issues.apache.org/jira/browse/SOLR-9555
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Alan Woodward
>
> See
> https://jenkins.thetaphi.de/job/Lucene-Solr-master-Linux/17888/consoleFull
> for an example
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
