[
https://issues.apache.org/jira/browse/HBASE-21828?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16758936#comment-16758936
]
Duo Zhang commented on HBASE-21828:
-----------------------------------
{quote}
I can see you mentioned some problems with CompletionException on HBASE-21779,
was that due nested CompletableFuture.whenComplete calls?
{quote}
Yes, but the effect is not only for HBASE-21779.
By design, if there is a TableNotFoundException which is a
DoNotRetryIOException, we will fail immediately. But the exception will be
wrapped by a CompletionException, as in the withTimeout method we return the
future which is generated by a whenComplete call, so the fail recovery logic
does not work and we will keep retrying until we reach any limit(the max retry
number or time).
And for HBASE-21779, the upper layer will also see a CompletionException
instead of TableNotFoundException, so the 'skip wal entries for deleted table'
logic is also broken, and cause some failures for UTs.
> Make sure we do not return CompletionException when locating region
> -------------------------------------------------------------------
>
> Key: HBASE-21828
> URL: https://issues.apache.org/jira/browse/HBASE-21828
> Project: HBase
> Issue Type: Sub-task
> Components: asyncclient, Client
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Priority: Major
> Fix For: 3.0.0, 2.2.0, 2.1.3, 2.0.5, 2.3.0
>
> Attachments: HBASE-21828.patch
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
