hi, Emmanuel:
If you are using glusterfs 3.3, this patch http://review.gluster.com/#patch,sidebyside,362,16,rpc/rpc-transport/socket/src/socket.c
and the following code in the patch will be helpful:
if (priv->own_thread) {
/*
* Without this, reconnect (= disconnect + connect)
* won't work except by accident.
*/
close(priv->sock);
priv->sock = -1;
++(priv->socket_gen);
}
Take a try, good luck.
Best Regards.
Jules Wang.
At 2012-09-25 12:35:16,"Emmanuel Dreyfus" <manu@xxxxxxxxxx> wrote:
>Emmanuel Dreyfus <manu@xxxxxxxxxx> wrote:
>
>> That bug is not that isolated. I reproduced it after a few days of
>> building like explained below.
>
>And I got it again, when accessing the same file. It takes hours, but it
>seems I can reproduce it.
>
>--
>Emmanuel Dreyfus
>http://hcpnet.free.fr/pubz
>manu@xxxxxxxxxx
>
>_______________________________________________
>Gluster-devel mailing list
>Gluster-devel@xxxxxxxxxx
>https://lists.nongnu.org/mailman/listinfo/gluster-devel
