systemd-oomd.service(8)

1SYSTEMD-OOMD.SERVICE(8)      systemd-oomd.service      SYSTEMD-OOMD.SERVICE(8)
2
3
4

NAME

6       systemd-oomd.service, systemd-oomd - A userspace out-of-memory (OOM)
7       killer
8

SYNOPSIS

10       systemd-oomd.service
11
12       /usr/lib/systemd/systemd-oomd
13

DESCRIPTION

15       systemd-oomd is a system service that uses cgroups-v2 and pressure
16       stall information (PSI) to monitor and take corrective action before an
17       OOM occurs in the kernel space.
18
19       You can enable monitoring and actions on units by setting
20       ManagedOOMSwap= and ManagedOOMMemoryPressure= in the unit
21       configuration, see systemd.resource-control(5).  systemd-oomd retrieves
22       information about such units from systemd(1) when it starts and watches
23       for subsequent changes.
24
25       Cgroups of units with ManagedOOMSwap= or ManagedOOMMemoryPressure= set
26       to kill will be monitored.  systemd-oomd periodically polls PSI
27       statistics for the system and those cgroups to decide when to take
28       action. If the configured limits are exceeded, systemd-oomd will select
29       a cgroup to terminate, and send SIGKILL to all processes in it. Note
30       that only descendant cgroups are eligible candidates for killing; the
31       unit with its property set to kill is not a candidate (unless one of
32       its ancestors set their property to kill). Also only leaf cgroups and
33       cgroups with memory.oom.group set to 1 are eligible candidates; see
34       OOMPolicy= in systemd.service(5).
35
36       oomctl(1) can be used to list monitored cgroups and pressure
37       information.
38
39       See oomd.conf(5) for more information about the configuration of this
40       service.
41

SYSTEM REQUIREMENTS AND CONFIGURATION

43       The system must be running systemd with a full unified cgroup hierarchy
44       for the expected cgroups-v2 features. Furthermore, memory accounting
45       must be turned on for all units monitored by systemd-oomd. The easiest
46       way to turn on memory accounting is by ensuring the value for
47       DefaultMemoryAccounting= is set to true in systemd-system.conf(5).
48
49       The kernel must be compiled with PSI support. This is available in
50       Linux 4.20 and above.
51
52       It is highly recommended for the system to have swap enabled for
53       systemd-oomd to function optimally. With swap enabled, the system
54       spends enough time swapping pages to let systemd-oomd react. Without
55       swap, the system enters a livelocked state much more quickly and may
56       prevent systemd-oomd from responding in a reasonable amount of time.
57       See "In defence of swap: common misconceptions"[1] for more details on
58       swap. Any swap-based actions on systems without swap will be ignored.
59       While systemd-oomd can perform pressure-based actions on such a system,
60       the pressure increases will be more abrupt and may require more tuning
61       to get the desired thresholds and behavior.
62
63       Be aware that if you intend to enable monitoring and actions on
64       user.slice, user-$UID.slice, or their ancestor cgroups, it is highly
65       recommended that your programs be managed by the systemd user manager
66       to prevent running too many processes under the same session scope (and
67       thus avoid a situation where memory intensive tasks trigger
68       systemd-oomd to kill everything under the cgroup). If you're using a
69       desktop environment like GNOME or KDE, it already spawns many session
70       components with the systemd user manager.
71

USAGE RECOMMENDATIONS

73       ManagedOOMSwap= works with the system-wide swap values, so setting it
74       on the root slice -.slice, and allowing all descendant cgroups to be
75       eligible candidates may make the most sense.
76
77       ManagedOOMMemoryPressure= tends to work better on the cgroups below the
78       root slice. For units which tend to have processes that are less
79       latency sensitive (e.g.  system.slice), a higher limit like the default
80       of 60% may be acceptable, as those processes can usually ride out
81       slowdowns caused by lack of memory without serious consequences.
82       However, something like user@$UID.service may prefer a much lower value
83       like 40%.
84

OPTIONS

86       --dry-run
87           Do a dry run of systemd-oomd: when a kill is triggered, print it to
88           the log instead of killing the cgroup.
89
90       -h, --help
91           Print a short help text and exit.
92
93       --version
94           Print a short version string and exit.
95

NOTES

101        1. "In defence of swap: common misconceptions"
102           https://chrisdown.name/2018/01/02/in-defence-of-swap.html
103
104
105
106systemd 253                                            SYSTEMD-OOMD.SERVICE(8)