contrail-nodemgr leaves [contrail-cassan] zombie process

[Sergey Kreys]

contrail-nodemgr periodically calls contrail-cassandra-status utility,
via non-blocking subprocess.Popen() and does not take care about
its return code. As a result, when child process ends and gracefully
exits, it is still present in "ps" output as zombie process
"[contrail-cassan]":

root 1061 0.0 0.0 4464 700 ? Ss 09:37 0:00 /bin/sh -e /proc/self/fd/9
root 1064 0.1 0.0 62168 18460 ? S 09:37 0:04 \_ /usr/bin/python /usr/bin/supervisord --nodaemon -c /etc/contrail/supervisord_config.conf
root 1357 0.0 0.0 244600 32172 ? Sl 09:37 0:02 \_ python /usr/bin/contrail-nodemgr --nodetype=contrail-config
root 5644 0.1 0.0 0 0 ? Z 10:53 0:00 | \_ [contrail-cassan] <defunct>

It is not critical, as just one zombie process is present, and it is
actually exited. But since somebody may be scared and confused by this
"[contrail-cassan]", it is better just to not let this process become
zombie.

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/38417
Submitter: Sergey Kreys (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/38417
Committed: http://github.com/Juniper/contrail-controller/commit/d14806253f561c9c311cfb688afebb0b3427483b
Submitter: Zuul (<email address hidden>)
Branch: master

commit d14806253f561c9c311cfb688afebb0b3427483b
Author: Sergey Kreys <email address hidden>
Date: Mon Dec 18 13:47:51 2017 +0300

Fix to avoid contrail-cassandra-status zombie proc

Proposed fix: use "subprocess.call()" instead of "Popen()" for
"contrail-cassandra-status". Use it with "shell=True" - to avoid
zombies, and use "&" at the end - to run in non-blocking way.
And also do this for "contrail-cassandra-repair" as it has same problem.

Change-Id: Ibefde45932613bb2f600a049037b483a2d7b33a8
Closes-bug: 1738757

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/47568
Submitter: Sergey Kreys (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/48499
Submitter: Sergey Kreys (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/47568
Committed: http://github.com/Juniper/contrail-controller/commit/29b8c08f328281dce89fe2fd009920f907bcb3e2
Submitter: Zuul (<email address hidden>)
Branch: R3.2

commit 29b8c08f328281dce89fe2fd009920f907bcb3e2
Author: Sergey Kreys <email address hidden>
Date: Mon Dec 18 13:47:51 2017 +0300

Fix to avoid contrail-cassandra-status zombie proc

Proposed fix: use "subprocess.call()" instead of "Popen()" for
"contrail-cassandra-status". Use it with "shell=True" - to avoid
zombies, and use "&" at the end - to run in non-blocking way.
And also do this for "contrail-cassandra-repair" as it has same problem.

Change-Id: Ibefde45932613bb2f600a049037b483a2d7b33a8
Closes-bug: 1738757

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/48499
Committed: http://github.com/Juniper/contrail-controller/commit/a1419a689e9690a02daf5587207e2c8647bd7d91
Submitter: Zuul (<email address hidden>)
Branch: R4.1

commit a1419a689e9690a02daf5587207e2c8647bd7d91
Author: Sergey Kreys <email address hidden>
Date: Mon Dec 18 13:47:51 2017 +0300

Fix to avoid contrail-cassandra-status zombie proc

Proposed fix: use "subprocess.call()" instead of "Popen()" for
"contrail-cassandra-status". Use it with "shell=True" - to avoid
zombies, and use "&" at the end - to run in non-blocking way.
And also do this for "contrail-cassandra-repair" as it has same problem.

Change-Id: Ibefde45932613bb2f600a049037b483a2d7b33a8
Closes-bug: 1738757