[
https://issues.apache.org/jira/browse/HBASE-28637?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Wei-Chiu Chuang updated HBASE-28637:
------------------------------------
Description:
AsyncFSWAL does not verify that WAL gets closed properly.
For example, HDDS-10609 found that Ozone close may fail because it runs out of
space. In such a case, the executor thread responsible for closing WAL file
simply exits, and the WAL remains open.
It is okay for HDFS WAL to stay open, but Ozone does not support renaming open
files, and therefore WAL archiving failed.
See HDDS-10609 for more details.
was:
AsyncFSWAL does not verify that WAL gets closed properly.
For example, HDDS-10609 found that Ozone close may fail because it runs out of
space. In such a case, the executor thread responsible for closing WAL file
simply exits, and the WAL remains open.
It is okay for HDFS WAL to stay open, but Ozone does not support renaming
directories that has open files, and therefore WAL archiving failed.
See HDDS-10609 for more details.
> asyncwal should attempt to recover lease if close fails
> -------------------------------------------------------
>
> Key: HBASE-28637
> URL: https://issues.apache.org/jira/browse/HBASE-28637
> Project: HBase
> Issue Type: Bug
> Reporter: Wei-Chiu Chuang
> Assignee: Wei-Chiu Chuang
> Priority: Major
>
> AsyncFSWAL does not verify that WAL gets closed properly.
> For example, HDDS-10609 found that Ozone close may fail because it runs out
> of space. In such a case, the executor thread responsible for closing WAL
> file simply exits, and the WAL remains open.
> It is okay for HDFS WAL to stay open, but Ozone does not support renaming
> open files, and therefore WAL archiving failed.
> See HDDS-10609 for more details.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
