Pacemaker logs contain errors from 'umm' utility: 'p_rabbitmq-server_start_0:22168:stderr [ /usr/bin/umm: line 40: runlevel: command not found ]'
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Released
|
Low
|
Artem Panchenko | ||
8.0.x |
Fix Released
|
Low
|
Anton Chevychalov |
Bug Description
While debugging an issue with RabbitMQ server I noticed that pacemaker logs contains a lot of such lines:
fgrep umm /var/log/
Dec 22 20:31:37 [18915] node-1.
Dec 22 20:31:37 [18915] node-1.
Dec 22 20:31:37 [18915] node-1.
fgrep -c umm /var/log/
5046
There is a script which is executed when BASH profile is loaded on nodes:
root@node-1:~# cat /etc/profile.
[ "z`umm status`" == "zumm" ] && cat /etc/issue.mm
And each time ocf script executes commands as 'rabbitmq' user [0] it gets an error in stderr from bash, like this one:
root@node-1:~# su - rabbitmq -s /bin/bash -c 'whoami'
/usr/bin/umm: line 40: runlevel: command not found
rabbitmq
It happens because 'runlevel' binary is not in PATH:
root@node-1:~# which runlevel
/sbin/runlevel
root@node-1:~# su - rabbitmq -s /bin/bash -c 'echo $PATH'
/usr/bin/umm: line 40: runlevel: command not found
/usr/local/
Steps to reproduce:
1. Deploy environment
2. Login to controller node via SSH
3. Execute command `su - rabbitmq -s /bin/bash -c 'whoami'`
Expected result: it returns "rabbitmq" in stdout, stderr is empty
Actual: stderr contains error
[0] https:/
Changed in fuel: | |
milestone: | 8.0 → 9.0 |
tags: | added: customer-found |
tags: | added: ct1 |
tags: | added: on-verification |
Fix proposed to branch: master /review. openstack. org/260712
Review: https:/