[Gluster-devel] Client Timeout?

[Daniel Maher]

On Mon, 14 Apr 2008 14:54:30 +0200 "Dennis Kern"
<dennis.kern at gmail.com> wrote:

> I'm evaluate glusterfs and I setup an test-enviorment like in
> http://www.gluster.org/docs/index.php/Simple_High_Availability_Storage_with_GlusterFS_1.3#Simple_setup_.28_Only_AFR_.29
> described. But "only" with 2 insted of 3 machines. All of them are
> virtual machines (vmware)
> 
> I simulate a server crash on server "2" (disconnect network on
> vmware-console) at this point I'm not able to access to the mount
> point on the client side directly. I type ls directly after disconnect
> of server "2":
> 
> client:/mnt/gluster# time ls
> file1 file2 file3
> 
> real    3m21.505s
> user    0m0.004s
> sys     0m0.000s

Try setting the timeout for the NS and DS blocks to something low, like
10 or 20 seconds :

volume <ds>
  type protocol/client
  ...
  option transport-timeout 10   # <-- timeout in seconds
end-volume



-- 
Daniel Maher <dma AT witbe.net>