[jira] Commented: (HBASE-2492) "BindException: Cannot assign requested address" because all descriptors occupied in TIME_WAIT when random reading

["stack (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-2492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12878982#action_12878982 ] 

stack commented on HBASE-2492:
------------------------------

>From the list, subject "Re: experiences with hbase-2492"

{code}
Since I got no replies to my previous message (see below), I went ahead and set the tcp_tw_recycle to true. This worked like a charm. The number of sockets in TIME_WAIT went down from many thousands to just a couple (tens). Apparently, once set to true, the recycling happens quite eagerly. Most importantly, the regionservers no longer shut down (which was the goal). I am sharing the info here, just in case it might help someone sometime.
{code}

> "BindException: Cannot assign requested address" because all descriptors occupied in TIME_WAIT when random reading
> ------------------------------------------------------------------------------------------------------------------
>
>                 Key: HBASE-2492
>                 URL: https://issues.apache.org/jira/browse/HBASE-2492
>             Project: HBase
>          Issue Type: Bug
>    Affects Versions: 0.20.4, 0.20.5, 0.21.0
>            Reporter: stack
>             Fix For: 0.21.0
>
>
> See tail of hbase-2180 for detail on this issue.  If a lot of random reading, we start to us up descriptors because sockets stuck in TIME_WAIT.  Workaround described over in hbase-2180.  This issue covers fix for hdfs so we don't open a socket each pread.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.