[Gluster-devel] Daemon dying, I killed it! <sob>

[Jim Richardson]

On Sat, 2007-10-13 at 15:35 +0530, Anand Avati wrote:
> Jim,
>  can you please give a link to your spec files? 

Not sure what you mean by this, do you mean the gluster client and
server config files? 

> Also, the backtrace is not available in the pastebin link you sent in
> the previous mail.. can you kindly post it again?
> 
> thanks,
> avati
> 


Sorry, I must have set it to one day expunge.

Here's a more long lived version
 
http://glusterfs.pastebin.com/f1343ca74



There are two backtraces there, I was able to kill off the daemon twice,
easily, with bonnie64 running against it, if glusterfs was started in
DEBUG logging mode.



For further info, I have been running a 3 node cluster over the weekend,
with bonnie and siege running continuously, with no problems that I can
tell. This was with glusterfs running in WARNING log mode.  

> On 10/12/07, Jim Richardson <jrichardson at cardomain.com> wrote:
>         
>         Backtrace on pastebin:
>         
>         http://glusterfs.pastebin.com/d56a5adb0
>         
>         
>         
>         I have a 3 node setup, each node is both client and server. I
>         had run
>         glusterfs with -L DEBUG, and was running bonnie on one of the
>         nodes, 
>         while each node was writing and reading files through a perl
>         script.
>         
>         While running this set, one or another of the daemons on the
>         nodes *not*
>         running bonnie would die, pretty reliably if I gave it a few
>         minutes. 
>         
>         
>         I repeated this several times.
>         
>         I then ran glusterfs with -L WARNING and could not reliably
>         duplicate
>         it, although I did have one daemon die once, but I was doing a
>         couple
>         other things on that box at the time and haven't been able to
>         duplicate 
>         it.
>         
>         This is all with 1.3.5 of glusterfs, and the patched 2.7.0
>         fuse stuff,
>         on Centos5 w/linux 2.6.18-8.1.10 running on PIV SMP (HT)
>         boxes. More
>         details available if needed.
>         
>         
>         
>         --
>         Jim Richardson | CarDomain Network 
>         Sr. System Administrator
>         
>         jrichardson at cardomain.com | tel 206.926.2158 | cell
>         206.769-7327
>         http://www.cardomain.com/id/weaselkeeper
>         
>         
>         _______________________________________________
>         Gluster-devel mailing list
>         Gluster-devel at nongnu.org
>         http://lists.nongnu.org/mailman/listinfo/gluster-devel
> 
> 
> 
> -- 
> It always takes longer than you expect, even when you take into
> account Hofstadter's Law.
> 
> -- Hofstadter's Law
-- 
Jim Richardson | CarDomain Network
Sr. System Administrator

jrichardson at cardomain.com | tel 206.926.2158 | cell 206.769-7327
http://www.cardomain.com/id/weaselkeeper