Hello all, had an issue on with a virtualmachine on XenServer 7.0 yesterday. It is a Centos 7.4.1708 and in a OOM situation it just shut down. No kernel panic where I could see a message, but it just turned itself off. All I could see in the /var/log/messages was the following: ------------------------------------------------------------------------ Sep 24 03:11:08 hostname xe-daemon: Running CollectDisk ... Sep 24 03:11:21 hostname kernel: httpd invoked oom-killer: gfp_mask=0x280da, order=0, oom_score_adj=-1000 Sep 24 03:11:22 hostname kernel: httpd cpuset=/ mems_allowed=0 Sep 24 03:11:22 hostname kernel: CPU: 0 PID: 31557 Comm: httpd Not tainted 3.10.0-514.21.1.el7.x86_64 #1 Sep 24 03:11:22 hostname kernel: Hardware name: Xen HVM domU, BIOS 4.6.1-xs125300 05/09/2016 Sep 24 03:11:22 hostname kernel: ffff88000d9eb000 0000000023d5076e ffff8800120f3a78 ffffffff81686f13 Sep 24 03:11:22 hostname kernel: ffff8800120f3b08 ffffffff81681ebe ffffffff812aee5b ffff88003bf85910 Sep 24 03:11:22 hostname kernel: ffff88003bf85928 ffffffff00000206 fbfeefff00000000 0000000000000007 Sep 24 03:11:22 hostname kernel: Call Trace: Sep 24 03:11:22 hostname kernel: [<ffffffff81686f13>] dump_stack+0x19/0x1b Sep 24 03:11:22 hostname kernel: [<ffffffff81681ebe>] dump_header+0x8e/0x225 Sep 24 03:11:22 hostname kernel: [<ffffffff812aee5b>] ? cred_has_capability+0x6b/0x120 Sep 24 03:11:22 hostname kernel: [<ffffffff8113d203>] ? delayacct_end+0x73/0xb0 Sep 24 03:11:22 hostname kernel: [<ffffffff81184cfe>] oom_kill_process+0x24e/0x3c0 Sep 24 03:11:22 hostname kernel: [<ffffffff8118479d>] ? oom_unkillable_task+0xcd/0x120 ------------------------------------------------------------------------ Anybody seen this before? Is there a way to configure it so that it does a reboot instead of a shutdown? Greetings, Thomas
> On Sep 25, 2017, at 2:56 AM, Thomas Plant <thomas at plant.systems> wrote: > > OOM situationDo you have `sar` installed and activated? Does sar output show high memory leading up to the outage?
Am 25.09.2017 um 18:20 schrieb Steven Tardy:>> On Sep 25, 2017, at 2:56 AM, Thomas Plant <thomas at plant.systems> wrote: >> >> OOM situation > Do you have `sar` installed and activated? > Does sar output show high memory leading up to the outage? > _______________________________________________ > CentOS mailing list > CentOS at centos.org > https://lists.centos.org/mailman/listinfo/centosNo 'sar' installed. It is a client server where I do not want to mess around to much. Yesterday happened again, it tried to kill httpd processes, but all have a oom_score_adj of -1000. So nothing could be killed. Where does Centos 7.4 set the oom_score_adj? Maybe I should lower it, as it seems to me as there is mod_php involved and some processes get stuck and fill the memory. Greetings, Thomas