OpenVZ Forum


Home » General » Support » OpenVZ 7 OOM killing systemd in containers
OpenVZ 7 OOM killing systemd in containers [message #53709] Wed, 11 November 2020 09:59
noc.r is currently offline  noc.r
Messages: 8
Registered: July 2020
Junior Member
From: *pool80-102-34.dynamic.orange.es
We are experiencing some problems with OpenVZ 7 hosts killing systemd in containers, so the container is stopped.

Is there any way to limit/block/set this in order to prevent killing systemd when low on memory? Is this behaviour expected?
The loaded kernel is 3.10.0-1127.8.2.vz7.151.14

On OpenVZ 6 this could be configured in /etc/vz/oom-groups.conf , but on OpenVZ 7 seems that it should be configured for systemd services as explained in https://docs.virtuozzo.com/virtuozzo_hybrid_server_7_users_g uide/managing-resources/managing-containers-memory-parameter s.html . However, this would not apply to systemd itself.

This is an example of a systemd's container being killed:

oct 27 18:27:55 hostname kernel: Task in /machine.slice/3402 killed as a result of limit of /machine.slice/3402
oct 27 18:27:55 hostname kernel: memory: usage 8362832kB, limit 8388608kB, failcnt 1869533
oct 27 18:27:55 hostname kernel: memory+swap: usage 8388608kB, limit 8388608kB, failcnt 12306558
oct 27 18:27:55 hostname kernel: kmem: usage 570448kB, limit 4194304kB, failcnt 0
oct 27 18:27:55 hostname kernel: Memory cgroup stats for /machine.slice/3402: rss_huge:0KB mapped_file:10236KB shmem:726376KB slab_unreclaimable:219768KB swap:25776KB cache:739468KB rss:70
oct 27 18:27:55 hostname kernel: Memory cgroup out of memory: Kill process 644121 (systemd) score 20 or sacrifice child
oct 27 18:27:55 hostname kernel: Killed process 625032 (spamd) in VE "3402", UID 0, total-vm:246960kB, anon-rss:120712kB, file-rss:208kB, shmem-rss:0kB
oct 27 18:27:55 hostname kernel: php-cgi invoked oom-killer: gfp_mask=0xd0, order=0, oom_score_adj=0
oct 27 18:27:55 hostname kernel: php-cgi cpuset=3402 mems_allowed=0
oct 27 18:27:55 hostname kernel: CPU: 2 PID: 555524 Comm: php-cgi ve: 3402 Kdump: loaded Tainted: G        W  OE  ------------   3.10.0-1127.8.2.vz7.151.14 #1 151.14
oct 27 18:27:55 hostname kernel: Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./E3C246D4U2-2T, BIOS L2.02K 12/18/2019
oct 27 18:27:55 hostname kernel: Call Trace:
oct 27 18:27:55 hostname kernel:  [<ffffffffa87b67f1>] dump_stack+0x19/0x1b
oct 27 18:27:55 hostname kernel:  [<ffffffffa87b0fc6>] dump_header+0x90/0x229
oct 27 18:27:55 hostname kernel:  [<ffffffffa81d7076>] ? find_lock_task_mm+0x56/0xc0
oct 27 18:27:55 hostname kernel:  [<ffffffffa81d7dad>] oom_kill_process+0x47d/0x640
oct 27 18:27:55 hostname kernel:  [<ffffffffa82040fe>] ? get_task_oom_score_adj+0xee/0x100
oct 27 18:27:55 hostname kernel:  [<ffffffffa81d7213>] ? oom_badness+0x133/0x1e0
oct 27 18:27:55 hostname kernel:  [<ffffffffa825f509>] mem_cgroup_oom_synchronize+0x4b9/0x510
oct 27 18:27:55 hostname kernel:  [<ffffffffa81d84c3>] pagefault_out_of_memory+0x13/0x50
oct 27 18:27:55 hostname kernel:  [<ffffffffa87af06d>] mm_fault_error+0x6a/0x157
oct 27 18:27:55 hostname kernel:  [<ffffffffa87c49a1>] __do_page_fault+0x491/0x500
oct 27 18:27:55 hostname kernel:  [<ffffffffa87c4a45>] do_page_fault+0x35/0x90
oct 27 18:27:55 hostname kernel:  [<ffffffffa87c0778>] page_fault+0x28/0x30


Any help would be appreciated.
Previous Topic: OpenVZ 7 containers crashing with ext4 errors
Next Topic: BUG: "prlctl destroy <UUID>" autostarts containers
Goto Forum:
  


Current Time: Wed Dec 02 05:57:52 GMT 2020