[jira] [Resolved] (HBASE-27175) Failure to cleanup WAL split dir log should be at INFO level

["Viraj Jasani (Jira)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HBASE-27175?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Viraj Jasani resolved HBASE-27175.
----------------------------------
    Hadoop Flags: Reviewed
      Resolution: Fixed

> Failure to cleanup WAL split dir log should be at INFO level
> ------------------------------------------------------------
>
>                 Key: HBASE-27175
>                 URL: https://issues.apache.org/jira/browse/HBASE-27175
>             Project: HBase
>          Issue Type: Task
>            Reporter: Viraj Jasani
>            Assignee: Ujjawal Kumar
>            Priority: Minor
>             Fix For: 2.5.0, 3.0.0-alpha-4, 2.4.14
>
>
> As part of the SCP, after we are done splitting WALs, we try removing the -splitting dirs but if the dir doesn't exist, dfs#delete fails with IOE. Since we are aware of this case, we just log the message but we don't interrupt SCP because handling of the failure is gracefully done.
> Hence, failure to remove "-splitting" dir should not be mentioned as WARN log, let's convert this to INFO level:
> {code:java}
> LOG.warn("Remove WAL directory for {} failed, ignore...{}", serverName, e.getMessage()); {code}
>  
> Any other genuine failure to remove the splitting dir is anyways covered by this log, hence we don't have worry about keeping the above mentioned log at WARN level, we have anyways mentioned "ignore..." in the log message.
> {code:java}
> if (!fs.delete(splitDir, false)) {
>   LOG.warn("Failed delete {}, contains {}", splitDir, fs.listFiles(splitDir, true));
> } {code}



--
This message was sent by Atlassian Jira
(v8.20.10#820010)