[Sheepdog] [PATCH 0/2] fix collie command errors during node member changes

[Chris Webb]

MORITA Kazutaka <morita.kazutaka at lab.ntt.co.jp> writes:

> Probably, it is a bug of Sheepdog.  Is there an easy way to reproduce
> it with a small cluster?  I'd like to try to test it, too.

Hi Kazutaka. I just started a small cluster of three machines (with three
sheep per machine on three different drives, but I'm sure it would work just
as well with only one), did a cluster format with --copies=2, and wrote a
vdi to the cluster so I had something to test with.

I then (effectively---actually did an ip link set ethX down) unplugged the
network to one of the machines. When I did a collie vdi list on one of the
machines in the remaining cluster, it paused until it noticed the machine
had gone, then continued correctly. However, the sheep daemon never seemed
to exit on the machine that had been disconnected, and collie vdi list just
hung forever. It seems to happen this way every time so is probably very
easy to reproduce.

Cheers,

Chris.