accumulo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Keith Turner <ke...@deenlo.com>
Subject Re: Table deletion got stuck
Date Wed, 28 Nov 2012 16:56:41 GMT
On Tue, Nov 27, 2012 at 4:28 PM, Lin XIAO <jackie.pku@gmail.com> wrote:

> When deleting tables, sometimes the operation get stuck for hours with
> table status "UNKNOWN" shown in the monitor page. The only relevant
> logs I can find are similar to this:
>
> [tableOps.CleanUp] DEBUG: Still waiting for table to be deleted: is
> locationState: is<<@(null,10.0.0.10:41000[43b1b039a081368],null)
>
> It happens to a few tables in our cluster and most of them are hosted
> by one tablet server, so I killed the tablet server to force the
> tablets to be reassigned to other servers. The tables got deleted
>

In the future save the output of running jstack on the tablet server before
killing it.


> successfully after that. I'm wondering what I should do to solve
> problems like this. Right now, we still have a bunch of tables with
> status "UNKNOWN". I'm wondering if I need to find the tablet servers
> hosting those tables and restart them.
>
> Btw, the swappiness in our cluster is set to 60 now instead of the
> recommended value 10.
>
> Thanks,
> Lin
>

Mime
View raw message