[
https://issues.apache.org/jira/browse/SOLR-8227?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14999928#comment-14999928
]
Ayon Sinha commented on SOLR-8227:
----------------------------------
Hi [[email protected]],
We have found that "the leader starts forwarding updates to the replica" is
extremely bad for the cluster. In a large cluster, when even one replica goes
into recovery due to heavy load, the leader simply waits forever for an ack
from that replica for every update and the updates back up very fast.
Due to SolrCloud following a strict CP model where every update has to be
ack-ed by every replica before leader calls it done, recovering from a leader
or replica must be identical in behavior.
Let me find the JIRA for the leader-to-recovering-replica that either
[~varunthacker] or [~tpot] has filed.
> Recovering replicas should be able to recover from any active replica
> ---------------------------------------------------------------------
>
> Key: SOLR-8227
> URL: https://issues.apache.org/jira/browse/SOLR-8227
> Project: Solr
> Issue Type: Improvement
> Reporter: Varun Thacker
>
> Currently when a replica goes into recovery it uses the leader to recover. It
> first tries to do a PeerSync. If thats not successful it does a
> replication. Most of the times it ends up doing a full replication because
> segment merging, autoCommits causing segments to be formed differently on the
> replicas ( We should explore improving that in another issue ) .
> But when many replicas are recovering and hitting the leader, the leader can
> become a bottleneck. Since Solr is a CP system , we should be able to recover
> from any of the 'active' replicas instead of just the leader.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
