[sheepdog] cluster_recovery_completion function bug

Liu Yuan namei.unix at gmail.com
Fri Mar 15 08:16:18 CET 2013


On 03/15/2013 02:48 PM, 王帆 wrote:
> for example.If the the node1 disk size is 1G,node2 is 2G node3 is
> 3G. sd_store->cleanup() will not be executed forever.

It seems that we indeed have problems,

--------------------------
mkdir store
for i in `seq 0 2`; do
	sheep/sheep  -n -z $i -p $((7000+$i)) -d -c local store/$i -s $(($i+1))000
done
sleep 1
collie/collie cluster format
sleep 1
collie/collie vdi create test 100M -P
sheep/sheep  -n -z 3 -p 7003 -d -c local store/3 -s 4000
ls store/*/obj/.stale
--------------------------

Above script won't function as expected. It would nice of you to come up with a patch to address
this problem and add a test script in tests/.

Thanks,
Yuan




More information about the sheepdog mailing list