upgrade_storage abort on service cinder-api restart failure

[wenqing liang]

upgrade_storage abort on service cinder-api restart failure upgrading from beta ML-2711 to R3.0-2717 kilo. Something may be wrong with upgrade_contrail since a few other services are not recognized, something to do with supervisord configs...

root@cmbu-vse2100-1:/etc/init# service nova-api restart
nova-api: unrecognized service
root@cmbu-vse2100-1:/etc/init# service nova-scheduler restart
nova-scheduler: unrecognized service

2016-02-24 17:22:14:207546: [root@10.87.129.173] out: [10.87.129.173] out: [localhost] local: sudo service cinder-api restart
2016-02-24 17:22:14:215108: [root@10.87.129.173] out: [10.87.129.173] out: cinder-api: unrecognized service
2016-02-24 17:22:14:230774: [root@10.87.129.173] out: [10.87.129.173] out:
2016-02-24 17:22:14:230930: [root@10.87.129.173] out: [10.87.129.173] out: Fatal error: local() encountered an error (return code 1) while executing 'sudo service cinder-api restart'
2016-02-24 17:22:14:231028: [root@10.87.129.173] out: [10.87.129.173] out:
2016-02-24 17:22:14:231123: [root@10.87.129.173] out: [10.87.129.173] out: Aborting.
2016-02-24 17:22:14:231247: [root@10.87.129.173] out: [10.87.129.173] out:
2016-02-24 17:22:14:395487: [root@10.87.129.173] out:
2016-02-24 17:22:14:427277: [root@10.87.129.173] out:
2016-02-24 17:22:14:427407: [root@10.87.129.173] out: Fatal error: run() received nonzero return code 1 while executing!
2016-02-24 17:22:14:427502: [root@10.87.129.173] out:
2016-02-24 17:22:14:427598: [root@10.87.129.173] out: Requested: sudo storage-fs-setup --storage-master 10.87.129.173 --storage-setup-mode upgrade --storage-hostnames cmbu-vse2100-1 cmbu-vse2100-10 cmbu-vse2100-11 --storage-compute-hostnames cmbu-vse2100-10 cmbu-vse2100-11 --storage-hosts 10.87.129.173 10.87.129.174 10.87.129.175 --storage-host-tokens n1keenA n1keenA n1keenA --storage-disk-config cmbu-vse2100-10:/dev/sdc cmbu-vse2100-11:/dev/sdc --storage-ssd-disk-config none --storage-journal-config cmbu-vse2100-10:/dev/sdb cmbu-vse2100-11:/dev/sdb --storage-local-disk-config none --storage-local-ssd-disk-config none --storage-nfs-disk-config none --storage-directory-config none --storage-chassis-config none --storage-mon-hosts none --collector-hosts 10.87.129.173 --collector-host-tokens n1keenA --cfg-host 10.87.129.173 --cinder-vip none --config-hosts none --storage-os-hosts none --storage-os-host-tokens none --cfg-vip none --storage-replica-size None --openstack-ip 10.87.129.173 --orig-hostnames cmbu-vse2100-1 cmbu-vse2100-10 cmbu-vse2100-11 --service-dbpass c0ntrail123 --region-name RegionOne

[Ignatious Johnson Christopher (ijohnson-x)]

Issue:
-----------
During upgrade the postinst of contrail-openstack back's up the
/etc/init.d/<openstack service> assuming that the conf file is from the
native package.
 apt-get install during upgrade wont unpack the conf files, if there is no change in configuration files( even with --force-confnew) option

Though the services are running, no service operation on individual service is possible as the init.d script is backed up to /etc/init.d/<service>.backup

root@cmbu-vse2100-1:/opt/contrail/utils# supervisorctl -s unix:///tmp/supervisord_openstack.sock status
cinder-api RUNNING pid 4715, uptime 17:21:27
cinder-scheduler RUNNING pid 4725, uptime 17:21:27
glance-api RUNNING pid 4714, uptime 17:21:27
glance-registry RUNNING pid 4723, uptime 17:21:27
heat-api RUNNING pid 4724, uptime 17:21:27
heat-api-cfn RUNNING pid 4720, uptime 17:21:27
heat-engine RUNNING pid 4722, uptime 17:21:27
keystone RUNNING pid 4718, uptime 17:21:27
nova-api RUNNING pid 4712, uptime 17:21:27
nova-conductor RUNNING pid 4713, uptime 17:21:27
nova-console RUNNING pid 4719, uptime 17:21:27
nova-consoleauth RUNNING pid 4716, uptime 17:21:27
nova-novncproxy RUNNING pid 4711, uptime 17:21:27
nova-objectstore RUNNING pid 4717, uptime 17:21:27
nova-scheduler RUNNING pid 4721, uptime 17:21:27
root@cmbu-vse2100-1:/opt/contrail/utils#

Fix:
-------
Fix is to backup only if the native package upstart
file exists(meaning fresh install). If the native package upstart dosent exists
it is upgrade case.

Workaround:
---------------------
Workaround is to rename /etc/init.d/service.backup to /etc/init.d/service

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/17839
Submitter: Ignatious Johnson Christopher (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Review in progress for https://review.opencontrail.org/17840
Submitter: Ignatious Johnson Christopher (<email address hidden>)

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/17840
Committed: http://github.org/Juniper/contrail-packaging/commit/9f14f0c0301a5bdd442cbf5f1022724f4d80a7f3
Submitter: Zuul
Branch: master

commit 9f14f0c0301a5bdd442cbf5f1022724f4d80a7f3
Author: Ignatious Johnson Christopher <email address hidden>
Date: Thu Feb 25 10:32:27 2016 -0800

During upgrade the postinst of contrail-openstack back's up the
/etc/init.d/<openstack service> assuming that the conf file is from the
native package. Fix is to backup only if the native package upstart
file exists(meaning fresh install). If the native package upstart dosent exists
it is upgrade case.

Change-Id: I27546002e0a411285bc98ac029d8fc26d1b8b38e
Closes-Bug: 1549619

[OpenContrail Admin (ci-admin-f)]

Reviewed: https://review.opencontrail.org/17839
Committed: http://github.org/Juniper/contrail-packaging/commit/2e6cb74b651a37bb05a507c86b5344b53b1569ef
Submitter: Zuul
Branch: R3.0

commit 2e6cb74b651a37bb05a507c86b5344b53b1569ef
Author: Ignatious Johnson Christopher <email address hidden>
Date: Thu Feb 25 10:32:27 2016 -0800

During upgrade the postinst of contrail-openstack back's up the
/etc/init.d/<openstack service> assuming that the conf file is from the
native package. Fix is to backup only if the native package upstart
file exists(meaning fresh install). If the native package upstart dosent exists
it is upgrade case.

Change-Id: I27546002e0a411285bc98ac029d8fc26d1b8b38e
Closes-Bug: 1549619