View Issue Details
|ID||Project||Category||View Status||Date Submitted||Last Update|
|0017708||CentOS-7||systemd||public||2020-09-02 07:08||2020-09-02 07:23|
|Summary||0017708: Top shows that the systemd-logind and dbus-daemon processes occupy nearly 100% of the CPU resources|
|Description||systemd version the issue has been seen with|
Expected behaviour you didn't see
system-logind and dbus should not occupy one cpu nearly 100% and there should be no watchdog behavior in system-logind（log as "systemd-logind.service watchdog timeout (limit 1min)!" ）
Unexpected behaviour you saw
huge closing sessions almost 30w+ appeared in the /run/systemd/sessions directory
Steps to reproduce the problem
the issue machines have the huge multi-sessions connections from remote login via ssh and crond tasks in localhost with 5min cycles. and I could not find the stable approach to reproduce the issue until now:(
I searched the issue with google and centos bugzilla, then found that the issue is a know issue as below link:
the issue has been resolved and needs to update the systemd version to solve the issue, but I could not update the systemd to the newest version because the fixup patches not related to the issue are so much, must consider compatibility and avoid bringing new issue for centos 7.4.
We have so much online machine using systemd 219.42.el7_4.4, update to newest is difficult for us.
The best way for us is to backport the fixup patches to the 219.42.el7_4.4, So my question is could you please help me:
1. provide us the fixup patches only for solving the above issue, we can backport the patches to our issue version?
2. provide us the stable approach to reproduce this issue?
btw, we think the issue is not same with #1961
we had checked the above patches which have been merged into 219.42.el7_4.4 version
Thanks for taking the time～
|Steps To Reproduce||the issue machines have the huge multi-sessions connections from remote login via ssh and crond tasks in localhost with 5min cycles. and I could not find the stable approach to reproduce the issue until now:(|
|Tags||No tags attached.|
for the above issue,
I had found four maybe fixup patches as below which almost have relationship with the issue.
Could you please help me make sure whether these patches are valid to the issue?)
if these are right, we will backport to our systemd/dbus version.
- core: use an AF_UNIX/SOCK_DGRAM socket for cgroup agent notification (d8fdc62)
- Only read one message at a time if there are fds pending (892f084eeda0)
- bus: Fix timeout restarts (529600397bca)
- DBusMainLoop: ensure all required timeouts are restarted (446b0d9ac75a)
CentOS 7.4 is ancient and unsupported. Run `yum update` and get yourself onto 7.8 which is 3 whole years newer and systemd-219-73.el7_8.9.x86_64 vs the ancient version you have now. There are over 400 lines in the rpm changelog from 219-42 to 219-73.9 and any one of those could be the problem you describe.
And update more often than every 3 years.