systemvm: fix keepalived is always restarted when update config for monitor service#4386
systemvm: fix keepalived is always restarted when update config for monitor service#4386ustcweizhou wants to merge 1 commit into
Conversation
…onitor service in 4.15, keepalievd in redundant VRs keeps restarting every minute. After debugging, I found it happens when update config for monitor service. it is because keepalived process is changed in Debian 10. in Debian 9 (systemvm for 4.14), ``` root@r-1969-VM:~# ps -ef|grep keepalived root 16324 1 0 09:53 ? 00:00:04 /usr/sbin/keepalived root 16325 16324 0 09:53 ? 00:00:04 /usr/sbin/keepalived root 16326 16324 0 09:53 ? 00:00:14 /usr/sbin/keepalived ``` in Debian 10 (systemvm for 4.15), processes end with "--dont-fork" ``` root@r-2040-VM:~# ps -ef|grep keepalived root 5237 1 0 16:40 ? 00:00:00 /usr/sbin/keepalived --dont-fork root 5239 5237 0 16:40 ? 00:00:03 /usr/sbin/keepalived --dont-fork ```
| @@ -194,7 +194,7 @@ def _redundant_on(self): | |||
| heartbeat_cron.commit() | |||
|
|
|||
| proc = CsProcess(['/usr/sbin/keepalived']) | |||
There was a problem hiding this comment.
@ustcweizhou specifying "/usr/sbin/keepalived" here may not be required as grep is used instead.
I understand this issue is with keepalived process, but on a broader view, other processes might have similar issue later, can this be fixed in CsProcess.py find() ?
There was a problem hiding this comment.
@sureshanaparti yes, of course.
To be sure that is not regression issue, I choosed a easier way to fix it.
There was a problem hiding this comment.
ok @weizhouapache , got it. Note that, if the same issue repeats for other processes later, then CsProcess.py have to be fixed.
|
@rhtyd @DaanHoogland this needs to be merged asap. the test failure should be fixed by this pr. Test | Result | Time (s) | Test File |
|
ok @weizhouapache , will run tests once |
|
@DaanHoogland a Jenkins job has been kicked to build packages. I'll keep you posted as I make progress. |
|
Packaging result: ✔centos7 ✔centos8 ✔debian. JID-2158 |
|
@blueorangutan test |
|
@DaanHoogland a Trillian-Jenkins test job (centos7 mgmt + kvm-centos7) has been kicked to run smoke tests |
|
Trillian test result (tid-2930)
|
|
none of these errors are related, @rhtyd @sureshanaparti or others, can we merge? |
|
The same PR was sent to 4.14 #4384 - should we close this one and merge the other one? |
6 participants
Description
in 4.15, keepalived in redundant VRs keeps restarting every minute.
After debugging, I found it happens when update config for monitor service.
it is because keepalived process is changed in Debian 10.
in Debian 9 (systemvm for 4.14),
in Debian 10 (systemvm for 4.15), processes end with "--dont-fork"
Types of changes
Screenshots (if appropriate):
How Has This Been Tested?