2015-04-23 17:42 GMT+02:00 Marc Cousin <cousinm...@gmail.com>: > On 20/04/2015 11:51, Marc Cousin wrote: >> On 31/03/2015 19:05, David Sterba wrote: >>> On Mon, Mar 30, 2015 at 05:09:52PM +0200, Marc Cousin wrote: >>>>> So it would be good to sample the active threads and see where it's >>>>> spending the time. It could be the somewhere in the rb-tree representing >>>>> extents, but that's a guess. >>>>> >>>> I just need to be told how to do that :) >>>> >>>> Something like a perf top ? >>> >>> for i in `pgrep btrfs`; do >>> ps $i >>> sudo cat /proc/$i/stack 2>/dev/null >>> done >>> >>> and capture samples every second. >>> >> >> Sorry, it took me a while to reproduce this (I needed to be in front of the >> machine when it happens :) ) > > Hi, > > Anything more I can provide on this issue ? > > Regards >
Hi, It took me a while (two years :) ), but I think I found what makees this problem appear, as I just triggered it on a system after this change. It occurs when enabling quotas on a volume. When there are a lot of snapshots that are deleted, the system becomes extremely unresponsive (IO often waiting for 30s on a SSD). When I don't have quotas, removing snapshots is fast. -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
