[SWARM][9.2] Failed to deploy env ceilometer ha with mongo

[Yury Tregubov]

Subj is failed from time to time.

The last one was on MOS 9.2 #438 snapshot
https://product-ci.infra.mirantis.net/job/9.x.system_test.ubuntu.services_ha.ceilometer/103/

The test is
Deploy cluster in ha mode with Ceilometer

        Scenario:
            1. Create cluster. Set install Ceilometer option
            2. Add 3 node with controller role
            3. Add 1 nodes with compute role
            4. Add 1 node with mongo role
            5. Deploy the cluster
            6. Verify **** api is running
            7. Run OSTF

        Duration 65m
        Snapshot: deploy_****_ha_with_mongo

One of the controllers went to error state:
[root@nailgun ~]# fuel nodes
id | status | name | cluster | ip | mac | roles | pending_roles | online | group_id
---+---------+---------------------+---------+------------+-------------------+------------+---------------+--------+---------
 4 | stopped | slave-03_controller | 1 | 10.109.5.5 | 64:72:64:0c:eb:91 | controller | | 1 | 1
 3 | stopped | slave-01_controller | 1 | 10.109.5.3 | 64:b3:21:d0:6c:41 | controller | | 1 | 1
 2 | stopped | slave-05_mongo | 1 | 10.109.5.7 | 64:f5:61:44:8c:46 | mongo | | 1 | 1
 1 | error | slave-02_controller | 1 | 10.109.5.4 | 64:ea:5f:90:fc:32 | controller | | 1 | 1
 5 | stopped | slave-04_compute | 1 | 10.109.5.6 | 64:4d:17:10:b0:d6 | compute | | 1 | 1

A lot of errors on the affected controller are seen from ceilometer.nova_client
ceilometer.agent.manager neutron.ml2.

[Sergii Rizvan (srizvan)]

The bug wasn't reproduced since 103rd run. Test 9.x.system_test.ubuntu.services_ha.ceilometer is failed from time to time but with different error. That's why status set as Invalid.