[ClusterLabs] Resource Agents: pgsql vs. pgsqlms

[Oyvind Albrigtsen]

On 11/07/24 22:20 GMT, Ede Wolf wrote:
>
>Once again thanks very much for your time.
>
>>>Just wondering, besides implementation and documentation, is there an
>>>overview of features / percieved advantages of one over the other?
>>
>>Not that I am aware of.
>
>That basically makes it even. In the sense, that from this point of 
>view there is no preference.
Yeah. In Red Hat we provide both, but I'm the maintainer of
the resource-agents upstream repository, so I got more experience with
that agent.

Using/improving both is of equal value to us.
>
>>My personal percieved advantages of pgsqlms over pgsql are:
>...
>
>>But pay attention on the fact that as the author of PAF, I'm highly biased, and
>>it's been ages since I had my hands on a cluster using the "pgsql" resource
>>agent. I don't really paid attention on how it evolved in the past 9 years.
>
>I am paying attention to the fact that the author of PAF replies to 
>beginners question. Beginner at least in this regard.
>
>Other than that, at least the copyright notice for the pgsql ra is 
>from 2012. Wether this is actually an indicates for any evolution I do 
>not know, as I have not done any further digging.
>It still should work, that's the message I am taking away from this 
>thread, with more or less feature parity, so we'll give it a try as 
>well.
That's just the last time the copyright notice for the agent was
updated. You can find the full list of commits at:
https://github.com/ClusterLabs/resource-agents/commits/main/heartbeat/pgsql

You'll find that the amount of commits decrease as the agent matures
(also in pgsqlms), unless there's some some feature added that
requires some improvements for a while.
>
>>>the other the language. Bash is realistic to read, perl probably not.
>>
>>That's really a matter of taste, and you might be surprised:
>
>Well, if we manage to get it up and running by help of documentation, 
>and some trial and error, we may not really need to read the agents 
>anyway. In fact, that is, what we hope. It would just be a fallback 
>since everyone here pitied either documentation
>
>>Too bad I hadn't update the vagrant PoC for a while, but you can have a look
>>in the following vagrant envs, if they still work with "modern" ansible/libvirt:
>
>For evaluation we actually prefer going completely on foot and not use 
>any automation magic. I am a bit oldfashioned and rather learn by 
>doing wrong than by reading yaml. What in turn prevents any further 
>career, but that is an entirely different story.
No automation is the best way to get started, as you should get used
to Pacemaker and the tools/agents, checking logs, "pcs resource
debug-<action> --full <rsc>" or setting trace_ra=1 for the resource
which will produce a file for each run of any action in
/var/lib/heartbeat/trace_ra/<rsc>.

That's also where you'll see the difference of bash vs perl, as bash
will show code line-by-line, which makes it easier to debug, but
also shows additional code for the log entries (as it shows all the
calls to show them on screen) vs perl where it will show debug-level
log messages in addition to the usual info/warn/error messages.

Automation will be nice to put on top of the basic skillset to
make deploying, monitoring, etc easier to repeat.
>
>Thanks again, very highly appreciated!
>
>Ede
>
>
>_______________________________________________
>Manage your subscription:
>https://lists.clusterlabs.org/mailman/listinfo/users
>
>ClusterLabs home: https://www.clusterlabs.org/
>