[ClusterLabs Developers] resource-agents v4.1.0 rc1

Ken Gaillot kgaillot at redhat.com
Mon Nov 20 15:26:51 UTC 2017


On Fri, 2017-11-17 at 17:04 +0100, Kristoffer Grönlund wrote:
> Dejan Muhamedagic <dejanmm at fastmail.fm> writes:
> 
> > > > > P.S. Right now irrelevant, but LVM-activate doesn't seem like
> > > > > a
> > > > > good name.
> > > > 
> > > > If nothing else, it's in keeping with the rest of the mess of
> > > > bad names
> > > > ;) Maybe we should decide on a naming standard? Abbreviations
> > > > or no
> > > > abbreviations, CamelCase or lowercase, or lower_case? This
> > > > could be
> > > > enforced through the CI checks, and the Makefile could make
> > > > sure to
> > > > install aliases to the old names so as not to break existing
> > > > installations.
> > > 
> > > I think we should keep them as is for current agents, but new
> > > agents
> > > should be lowercase with an optional -'s if that clarifies their
> > > usage.
> > 
> > My point was that it is arguably a pleonasm. The "activate" part
> > is superfluous. Sorry for not being clearer.
> > 
> > As for bad names, we need to look no further than the LVM RA:
> > in volgrpname the "name" part is just as much necessary.
> 
> While I agree that LVM-activate, being a verb, is not a great
> name, I at least am having a hard time coming up with a good one.
> Naming
> truly is the hardest part of programming...
> 
> I guess my general point was that we don't have any kind of
> guidelines
> for naming that people can refer to, resulting in the amazing variety
> of
> names both in style and semantics that we're stuck with now.
> 
> We had this discussion before and decided it was as much trouble as
> it
> solves to change things, but I still haven't given up on the idea ;)
> There's also the topic of the heartbeat provider being completely
> outdated and continuing to confuse newcomers, especially considering
> Pacemaker now dropping heartbeat support completely as of version
> 2. Maybe that would be the time to introduce a new provider, with a
> more
> strictly defined standard for agent and parameter naming for example.

Yes, it's frustrating sometimes how quickly time flies by :-)

I think the direction of the last discussion was that in the long-
planned overhaul of OCF, we would define some mechanism for
deprecating/aliasing agents, and take care of it then.

That's still the best approach, but a cheap version would be:

- rename the provider to "clusterlabs", rename all the agents as
desired, and leave out any obsolete agents

- create a "heartbeat" directory, put the obsolete agents there, and
symlink all the old names to the new clusterlabs agent names

e.g.
   /usr/lib/ocf/resource.d/
      clusterlabs/
         ip
      heartbeat/
         IPaddr2 -> ../clusterlabs/ip

The drawback compared to the planned approach is that the user wouldn't
be warned that they're using an old name.

> 
> > 
> > > > Maybe we could even move IPaddr2 to ip or IP, and move the now-
> > > > outdated
> > > > agents to a legacy provider...
> > 
> > Yeah, I hope we don't need to get into that ;-)
> 
> Too late! Anyway, I'll try to let it go for now... ;)
> 
> Cheers,
> Kristoffer
> 
> > 
> > Cheers,
> > 
> > Dejan
> > 
> > > > 
> > > > Cheers,
> > > > Kristoffer
> > > > 
> > > > > 
> > > > > 
> > > > > 
> > > > > 
> > > > > >  - minio
> > > > > >  - NodeUtilization
> > > > > >  - oraasm
> > > > > >  - ovsmonitor
> > > > > >  - rkt
> > > > > >  - ZFS
> > > > > > 
> > > > > > - bugfixes and enhancements:
> > > > > >  - aws*: fixes and improvements
> > > > > >  - CTDB: fixes for newer versions
> > > > > >  - CTDB: fix for --logfile being replaced with --logging
> > > > > >  - DB2: fix HADR support for DB2 V98+
> > > > > >  - docker: add docker-native healthcheck
> > > > > >  - galera: fix for MariaDB 10.1.21+
> > > > > >  - mysql: set correct master score after maintenance mode
> > > > > >  - ocf-shellfuncs: improve locking (ocf_take_lock())
> > > > > >  - pgsql: add support for PostgreSQL 10
> > > > > >  - pgsql: allow dynamic membership
> > > > > >  - rabbitmq-cluster: fix to work on Pacemaker remote nodes
> > > > > > 
> > > > > > The full list of changes for resource-agents is available
> > > > > > at:
> > > > > > https://github.com/ClusterLabs/resource-agents/blob/v4.1.0r
> > > > > > c1/ChangeLog
> > > > > > 
> > > > > > Everyone is encouraged to download and test the new release
> > > > > > candidate.
> > > > > > We do many regression tests and simulations, but we can't
> > > > > > cover all
> > > > > > possible use cases, so your feedback is important and
> > > > > > appreciated.
> > > > > > 
> > > > > > Many thanks to all the contributors to this release.
> > > > > > 
> > > > > > 
> > > > > > Best,
> > > > > > The resource-agents maintainers
-- 
Ken Gaillot <kgaillot at redhat.com>




More information about the Developers mailing list