Non quoted var in RabbitMQ OCF script leads to script failure

[Dmitry Mescheryakov]

On the 7.0 release the following error was observed in pacemaker logs:

Dec 16 22:49:09 [9351] node-1 pacemaker_remoted: notice: operation_finished: p_rabbitmq-server_notify_0:24211:stderr [ /usr/lib/ocf/resource.d/fuel/rabbitmq-server: line 1485: [: node-1: binary operator expected ]

here line 1485 corresponds to line 1530 in the current version of the script:
https://github.com/openstack/fuel-library/blob/ba38e4116c23a4b76c826221872d8bd614ea455b/files/fuel-ha-utils/ocf/rabbitmq#L1530

The issue happens when OCF_RESKEY_CRM_meta_notify_stop_uname contains name of more than one node. Which is a valid input according to Pacemaker docs, see Example 10.6 in
http://clusterlabs.org/doc/en-US/Pacemaker/1.1/html-single/Pacemaker_Explained/#_clone_resource_agent_requirements

There is no impact from that error at the moment: the if statement is just skipped and by chance that is the desired behaviour when several hosts are stopped. Report from user also indicates that the error does not cause any issue with RabbitMQ. Still, the importance is set to medium because the error might become dangerous after some change later.

[Dmitry Mescheryakov (dmitrymex)]

The issue is already fixed in 8.0, we just need to port the fix to 7.0

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/7.0
Review: https://review.openstack.org/259363

[Dmitry Mescheryakov (dmitrymex)]

Lowering importance to medium because there is no user impact for now.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/6.1
Review: https://review.openstack.org/273474

[OpenStack Infra (hudson-openstack)]

Change abandoned by Rodion Tikunov (<email address hidden>) on branch: stable/6.1
Review: https://review.openstack.org/273474
Reason: Will do for this cherry-pick a new commit

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/6.1
Review: https://review.openstack.org/275189

[Rodion Tikunov (rtikunov)]

Won't Fix for 6.1-updates because of Medium importance

[OpenStack Infra (hudson-openstack)]

Change abandoned by Fuel DevOps Robot (<email address hidden>) on branch: stable/6.1
Review: https://review.openstack.org/275189
Reason: This review is > 4 weeks without comment, and failed Jenkins the last time it was checked. We are abandoning this for now. Feel free to reactivate the review by pressing the restore button and leaving a 'recheck' comment to get fresh test results.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/259363
Committed: https://git.openstack.org/cgit/openstack/fuel-library/commit/?id=97d36d0bea8f22ab5029a2ddff0f5a49d0d28d23
Submitter: Jenkins
Branch: stable/7.0

commit 97d36d0bea8f22ab5029a2ddff0f5a49d0d28d23
Author: Dmitry Mescheryakov <email address hidden>
Date: Fri Dec 18 13:09:23 2015 +0300

    Quote vars when checking them on emptiness

    Not doing so leads to bash errors when string contains spaces.
    Specifically, variable OCF_RESKEY_CRM_meta_notify_stop_uname can
    contain spaces when several nodes are stopped simultaneously and
    that leads to error described in the bug referenced below.

    Regarding other variables quoted in this change "...promote_uname",
    "...master_uname" and "...demote_uname" - in our case they never
    should be empty, since we can have only one master. But better safe
    than sorry.

    Change-Id: I1447f58e897cc1fe6b9884c3b4cd4f950dd9cecb
    Closes-Bug: #1527518

[TatyanaGladysheva (tgladysheva)]

Verified on MOS 7.0 + MU4 updates.

Steps to verify:
1. Deploy 3 controllers+2 computes env
2. Open pacemaker log on every controller:
tail -f /var/log/pacemaker.log | grep p_rabbitmq-server_notify_0
3. Kill rabbitmq process (ps axuf | grep rabbitmq) one by one on different controllers and see output in logs from step 2. Repeat several times.

Actual result:
No errors like the following had been found in pacemaker logs after applying MU4 patch:
Jun 02 08:10:13 [18513] node-1.test.domain.local pacemaker_remoted: notice: operation_finished: p_rabbitmq-server_notify_0:25472:stderr [ /usr/lib/ocf/resource.d/fuel/rabbitmq-server: line 1543: [: node-3.test.domain.local: binary operator expected ]