test_server_numa in test_numa_topology doesn't work on Ubuntu 14
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
intel-nfv-ci-tests |
Fix Released
|
Undecided
|
Unassigned |
Bug Description
23:25:02 Captured traceback:
23:25:02 ~~~~~~~~~~~~~~~~~~~
23:25:02 Traceback (most recent call last):
23:25:02 File "tempest/test.py", line 113, in wrapper
23:25:02 return f(self, *func_args, **func_kwargs)
23:25:02 File "/usr/local/
23:25:02 placement = self.get_
23:25:02 File "/usr/local/
23:25:02 % out.strip(), shell=True)
23:25:02 File "/usr/local/
23:25:02 cmd=sanitized_cmd)
23:25:02 oslo_concurrenc
23:25:02 Command: grep name /proc/28256
23:25:02 28409/cgroup
23:25:02 Exit code: 127
23:25:02 Stdout: u''
23:25:02 Stderr: u'grep: /proc/28256: Is a directory\n/bin/sh: 2: 28409/cgroup: not found\n'
Reviewed: https:/ /review. openstack. org/309122 /git.openstack. org/cgit/ openstack/ intel-nfv- ci-tests/ commit/ ?id=824805d91dd 877ded1244dfd27 6fe59812621ead
Committed: https:/
Submitter: Jenkins
Branch: master
commit 824805d91dd877d ed1244dfd276fe5 9812621ead
Author: Waldemar Znoinski <email address hidden>
Date: Thu Apr 21 18:35:53 2016 +0100
fix get_placement function to use ps and new cgroup path
Change in path to cgroup that has vcpu placement
information is required for the function to work properly
when run on host VM with Ubuntu 14. This breaks this
function run in a docker container where cgroup paths
are different. get_placement function now uses
ps instead of pgrep due to fact pgrep was picking
up more than qemu/VM process - it was not filtering
'/bin/sh pgrep...' line.
Change-Id: I7772dee3f9cf86 80a65681fe642eb a6e0724f80e
Closes-Bug: #1573157