Vcenter-as-compute: SM provisioning rabbitmq-server shown inactive on contrail-status and stay across reboots

[Sarath]

Version: 4.0.1.0-69-mitaka also seen newton
Topology: 3node HA with multiple computes (multi-cluster Esxi) & Kvm

Contrail-status should either show correctly the state (or) it need not display as this kind of status mislead the customer

This issue seen on all of the controller nodes as below and stay same across reboots

root@5a10s31:~# docker exec -it controller bash
contrail-status
root@5a10s31(controller):/# contrail-status
exit
== Contrail Control ==
contrail-control: active
contrail-named: active
contrail-dns: active
contrail-control-nodemgr: active
== Contrail Config ==
contrail-api: active
contrail-schema: backup
contrail-svc-monitor: backup
contrail-device-manager: backup
contrail-config-nodemgr: active
== Contrail Config Database==
contrail-database: active

== Contrail Web UI ==
contrail-webui: active
contrail-webui-middleware: active
== Contrail Support Services ==
zookeeper: active
rabbitmq-server: inactive (disabled on boot)
root@5a10s31(controller):/# exit

[kamlesh parmar (kparmar)]

you are using rabbitmq on openstack, so controller will not have it. This is expected.

[Sarath (nsarath)]

From: Sarathbabu Narasimhan
Sent: Thursday, September 14, 2017 12:55 PM
To: Kamlesh Parmar <email address hidden>; Amudha R <email address hidden>; Sachchidanand Vaidya <email address hidden>; Sandip Dey <email address hidden>; Pavana Prakash <email address hidden>
Cc: Jeba Paulaiyan <email address hidden>; Sudheendra Rao <email address hidden>; Rudra Rugge <email address hidden>; Abhay Joshi <email address hidden>; Sarathbabu Narasimhan <email address hidden>
Subject: RE: 4.0.1 vCenter Testing

Hi Kamlesh,

As it is Usecase-1/topology and published documents showing it supported topology,
So we need to give “reliable output of current system/services states” for supported topologies
as contrail-status being used to check health of system/services.

As we control the “contrail-status” script can we find the way to get actual state of system/service (or)
can we include only what is possible to be displayed reliably so that it doesn’t mislead the customer
if they deploy this supported topology with our bundled software package.

Thanks
*Sarath

[Sarath (nsarath)]

RN: In usecase/topology with 3 node HA of both "Contrail + Openstack" roles played by same 3 nodes, rabbitmq-server actually working good but shown inactive being display-only issue.

[Nagendra Prasath (npchandran)]

contrail-status shows the necessary services expected to be running the controller node.
From above conversation i understand that rabbitmq is used from openstack node, and is actually disabled in the controller and hence you're seeing
"rabbitmq-server: inactive (disabled on boot)"
which i suppose would actually help understand that rabbitmq is disabled in contrail controller.

[Nagendra Prasath (npchandran)]

Updated description to help clarify the ambiguity... Let me know if this looks OK.

[Sarath (nsarath)]

As discussed offline, multiples of earlier and current TechPub refers to "provisioning success" of all components when they are active,
https://www.juniper.net/documentation/en_US/contrail4.0/topics/task/configuration/vcenter-integration-vnc.html

We need to correct them (or) add notes for this specific UseCase to set the expectations accordingly to this new changes of interpreting "inactive" as okay.

[Nagendra Prasath (npchandran)]

We can update techpub from R4.0 onwards. Would be difficult to correct each and everything. Have updated the doc provided in #6.

[Sarath (nsarath)]

Okay Nagendra, if Techpub from R4.0 onwards taken care should be good.

[Sachin Bansal (sbansal)]

Based on comment #8, I am closing the bug. If my understanding is incorrect, please reopen and provide details.