[sheepdog-users] [ANNOUNCE] sheepdog stable release v0.7.3-rc0

[Gerald Richter - ECOS]

Hi,

I have installed the patched version and didn't see any segfaults so far. Neither directly after formatting, nor when doing heaving writes into sheepdog, so it seems that the patch has fixed the problem.

Thanks for the quick fix

Regards

Gerald


At Mon, 09 Sep 2013 01:50:01 +0900,
Hitoshi Mitake wrote:
> 
> 
> At Sun, 8 Sep 2013 09:47:30 +0200,
> Gerald Richter - ECOS wrote:
> > 
> > Hi,
> > 
> > regarding the segfault I mentioned on Friday, there are two nodes, format was with copies = 3 and I am using corosync. I don't have much more information anymore, because I had to import data on the weekend. 
> > 
> > During the data import sheep crashed again. It crashed when I had two qemu-img convert running at the same time (of course with different vdi's) on the same machine. I still using 2 nodes. The network between the two node is only 100Mbit/s, so It got from time to time these poll timeout, but which wasn't a problem before. Here is the output of sheep.log (the second sheep on the machine were no import happen is still running).
> 
> I sent a patch which will be able to solve the below problem. Could you try it?
> # The patch will be pushed to the stable branch soon if other developers ack it.

Source archive which contains the above patch can be downloaded from here:
https://github.com/sheepdog/sheepdog/archive/stable-0.7.zip

Thanks,
Hitoshi