[
https://issues.apache.org/jira/browse/HBASE-29831?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18052060#comment-18052060
]
Kodey Converse commented on HBASE-29831:
----------------------------------------
Only affects major version 2, because replication for read replicas is handled
differently on 3
> Region replica's can fall behind due to a NPE during processing
> ---------------------------------------------------------------
>
> Key: HBASE-29831
> URL: https://issues.apache.org/jira/browse/HBASE-29831
> Project: HBase
> Issue Type: Bug
> Components: read replicas
> Reporter: Kodey Converse
> Priority: Minor
>
> The {{RegionReplicaReplicationEndpoint}} checks table descriptors while it's
> processing WAL edits to attempt to update it's cache of replication factors.
> This check does not handle a table being dropped while the endpoint is
> running (even if the table being dropped does not have replicas enabled),
> which results in the endpoint crashing and not recovering until a restart of
> the RegionServer. Replicas for all regions (including regions from other
> tables) hosted by the RegionServer stop being updated.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
