[ 
https://issues.apache.org/jira/browse/ACCUMULO-1416?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13657090#comment-13657090
 ] 

Keith Turner commented on ACCUMULO-1416:
----------------------------------------

This check was not related to walogs.  When the OS disk goes read only, it will 
leave this system in a really screwy state.  In this state existing connections 
are kept alive and new connections can not be formed.   So its possible to end 
with a live tserver that clients can not connect to.  

Changes were made for ACCUMULO-513 to make the master always get a new 
connection when getting tserver status (used to reuse connections).  This 
should help the master find and kill these zombie tservers.

It seems like w/ the changes in ACCUMULO-513 that this check could be removed.  
If not removed it should be made to focus on the OS disk, and not the data disk.
                
> FileSystemMonitor isn't necessary anymore?
> ------------------------------------------
>
>                 Key: ACCUMULO-1416
>                 URL: https://issues.apache.org/jira/browse/ACCUMULO-1416
>             Project: Accumulo
>          Issue Type: Bug
>          Components: tserver
>            Reporter: John Vines
>
> With the removal of direct walogs, do we still need tservers monitoring the 
> filesystem and dying if things go wonky? Just had it happen and the death of 
> the tserver didn't seem necessary, especially with multiple disks.
> Relevant stack trace-
> {code}
> 2013-05-14 08:54:55,693 [util.FileSystemMonitor] FATAL: Exception while 
> checking mount points, halting process
> java.lang.Exception: Filesystem /data/04 switched to read only
>         at 
> org.apache.accumulo.server.util.FileSystemMonitor.checkMounts(FileSystemMonitor.java:134)
>         at 
> org.apache.accumulo.server.util.FileSystemMonitor$1.run(FileSystemMonitor.java:91)
>         at java.util.TimerThread.mainLoop(Timer.java:534)
>         at java.util.TimerThread.run(Timer.java:484)
> {code}

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Reply via email to