On Mon, Jul 17, 2017, at 10:53 PM, Nigel Babu wrote:
On Mon, Jul 17, 2017, at 02:00 PM, Jeff Darcy wrote:On Mon, Jul 17, 2017, at 09:28 AM, Nigel Babu wrote:It appears that something changed in https://review.gluster.org/#/c/17771 . The tarball with the logs for that patch was 32G. That doesn't sound right. Is something writing to log excessively after this patch landed?I don't see anything in the patch itself, or in its immediate predecessors, that would account for this. However, it could be a less direct kind of effect. I'll take a look.Are you sure that's the right patch? I looked at the smoke log, which seems quite ordinary, and the regression log is pretty tiny because tests on release-3.8-fb don't actually do much. On the other hand, I don't see any other patches on my list that seem to be likely culprits either
Since the crashes are in brick daemons, I strongly suspect that the crashes are actually from https://review.gluster.org/#/c/17750/ (which does touch server code) rather than 17771 (which does not). I'll try to reproduce the problem here, and then work on a fix.
_______________________________________________ Gluster-devel mailing list Gluster-devel@xxxxxxxxxxx http://lists.gluster.org/mailman/listinfo/gluster-devel
