On Thu, Apr 30, 2009 at 12:44 PM, Andrew Beekhof <[email protected]> wrote:
>
> Putting <= 2.0.8 anywhere near a production environment is bordering on
> irresponsible.
It occurred to me later, that it would reasonable to ask in response:
If it was such a pile of garbage, wasn't releasing it in the first
place also irresponsible?
Sadly, in retrospect, the answer is "probably".
I did the best I could and each release was better than the last, but
the reality is that this wasn't a huge achievement.
The problem was that for something near the beginning of the software
life-cycle, the way Heartbeat was being "run" practically guaranteed
that those early releases of the crm were crap. Developers had zero
control over an erratic, spontaneous, release schedule and were given
little notice when one was about to take place.
Which meant as soon as we found out (usually by reading this list),
there was a mad scramble to finish whatever we were working on
(because no-one knew when then next one would be) and beat the rest
into something remotely acceptable. At which point the release
manager would disappear again.
Things got better (after 2.0.8) once I initiated use of the build
service to create regular "interim" releases, but the real turning
point was the Pacemaker split - when the distinction between stable
and development releases was also established.
The combination is what allowed the codebase to mature and stabilize
to the point where using a version from more than 3 months ago became
a sane thing to do.
This isn't just another case of a lazy developer who can't be bothered
supporting anything other than the current version.
I'm so vocal about having people upgrade because I know just how bad
those older versions are.
The thought of people actually running them scares the hell out of me.
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
