[ceilometer] Zombie processes of ceilometer-api on all controllers
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Mirantis OpenStack |
Won't Fix
|
Medium
|
Igor Degtiarov | ||
6.0.x |
Won't Fix
|
Medium
|
Igor Degtiarov | ||
8.0.x |
Won't Fix
|
Medium
|
Igor Degtiarov | ||
9.x |
Won't Fix
|
Medium
|
Igor Degtiarov |
Bug Description
Steps To Reproduce:
1. Download the latest MOS 8.0 build (#361 in my case)
2. Deploy environment with 3 controllers and 1 compute node, Ceilometer enabled, mongo nodes on controller nodes.
3. Login to controller node
4. Check count of ceilometer processes:
ps aux | grep -v grep | grep ceilometer-api
Observed Result:
We can see zombie processes on all controllers:
root@node-2:~# ps aux | grep -v grep | grep ceilometer-api
ceilome+ 18849 0.0 0.0 0 0 ? Z 13:40 0:00 [ceilometer-api] <defunct>
ceilome+ 30852 2.0 1.1 831828 58052 ? Ssl Dec29 31:40 /usr/bin/python /usr/bin/
Changed in mos: | |
assignee: | nobody → MOS Ceilometer (mos-ceilometer) |
milestone: | none → 8.0 |
tags: | added: ceilometer |
tags: |
added: area-ceilometer removed: ceilometer |
Changed in mos: | |
status: | Confirmed → Won't Fix |
tags: | added: customer-found support |
We found out that that bug is coming from werkzeug, which is creates zombie process if multiple workers run --- link https:/ /github. com/mitsuhiko/ werkzeug/ issues/ 810