I have a similar problem on 4.4.0-91. Happened two times so far on Oct 31st and Nov 30th, during the nightly dumps of a postgres database. Strange coincidence, it's the last day of the month in both cases, but we don't do anything special at that time compared to the rest of the month.
I'll attach the OOM errors from both incidents, seems to be a similar error to the one reported in this bug: gfp_mask=0x26000c0, order=2, oom_score_adj=0 The server is a dual CPU machine, so 2 NUMA nodes. We have disabled numa_balancing, zone_reclaim_mode, and transparent huge pages for performance reasons, they were causing the database to become unresponsive for a few minutes from time to time. We are now considering also running the database with numactl --interleave=all If needed I can try to run apport. ** Attachment added: "First OOM log" https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1734744/+attachment/5016589/+files/oom.log -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/1734744 Title: OOM killer with kernel 4.4.0-92 To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1734744/+subscriptions -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
