[ClusterLabs] can't create master/slave resource

[Tomas Jelinek]

Hi,

systemd resources cannot be used as master/slave resources. In order to 
use a resource as a master/slave, the resource must support promote and 
demote actions [1], which systemd resources don't.

# pcs resource create test systemd:postfix
# pcs resource master test
# pcs cluster verify -V
    error: native_unpack:        Resource test:0 is of type systemd and 
therefore cannot be used as a master/slave resource
    error: create_child_clone:   Failed unpacking resource test
    error: unpack_resources:     Failed unpacking master test-master
Errors found during check: config not valid

You need to use an ocf resource agent (with promote and demote actions 
implemented) for your resource for this to work.

Because the resource cannot be unpacked by pacemaker, it is not shown in 
the "pcs status" output - pcs doesn't get any info about it from 
pacemaker. This issue has been already discussed. Pacemaker will provide 
info when such errors occur so pcs will be able to display it. [2]

The resource may not be running but it is still defined in the 
configuration:
# pcs resource --full
  Master: test-master
   Resource: test (class=systemd type=postfix)
    Operations: monitor interval=60 timeout=100 (test-monitor-interval-60)
                start interval=0s timeout=100 (test-start-interval-0s)
                stop interval=0s timeout=100 (test-stop-interval-0s)

That's why you get an error that the id already exists.


I'm going to file a bug against pcs so it won't be possible to 
master/slave systemd resources as it is not supported anyway.

Regards,
Tomas

[1]: 
http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/html-single/Pacemaker_Explained/index.html#_requirements_for_multi_state_resource_agents
[2]: https://bugzilla.redhat.com/show_bug.cgi?id=1447951


Dne 19.9.2017 v 17:13 Tiemen Ruiten napsal(a):
> Hello,
> 
> We have a 3-node cluster (CentOS 7.4) with several systemd resources 
> configured. One of them I would like to create as a master/slave 
> resource, so following the RedHat documentation:
> 
> pcs resource create ivr systemd:japp at ivr - works, the service is started 
> on one of the nodes.
> pcs resource master ivr-master ivr - doesn't work as expected: the 
> service is stopped and the output of pcs resource show doesn't list it 
> anymore. However, if I try the command again, I get an error saying the 
> resource ivr already exists! I have to delete the resource and recreate 
> it to get the service to run.
> 
> pacemaker-libs-1.1.16-12.el7_4.2.x86_64
> pacemaker-cluster-libs-1.1.16-12.el7_4.2.x86_64
> pacemaker-1.1.16-12.el7_4.2.x86_64
> pacemaker-cli-1.1.16-12.el7_4.2.x86_64
> corosynclib-2.4.0-9.el7_4.2.x86_64
> corosync-2.4.0-9.el7_4.2.x86_64
> 
> Am I doing something wrong?
> 
> -- 
> Tiemen Ruiten
> Systems Engineer
> R&D Media
> 
> 
> _______________________________________________
> Users mailing list: Users at clusterlabs.org
> http://lists.clusterlabs.org/mailman/listinfo/users
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org
>