Mitani, are you out there?
So far, my main observation with pgCluster is that there doesn't seem to be
much leadership on it :( Looking at pgfoundry, there are three 'developers'
... one of them has already written it off ("3 years ago, I worked on PGCluster
for 2 months -- and I gave up") ... Mitani seems to be relatively unreachable,
either directly or through the lists, and the 3rd, I'm not sure of ...
My first experience with the latest version of PgCluster that Mitani put up in
Sep was a great pleasure ... it built/installed nicely, configuration was
literally brain dead ( a few holes in the docs that could be filled concerning
rsync and ssh keys, but beyond that, very straightforward) ...
Shortly after Mitani released 1.9 in Sep '09, Markus Schmidleitner posted a
patch:
http://lists.pgfoundry.org/pipermail/pgcluster-general/2009-October/002057.html
Big thing is that it brought pgCluster up to 8.3.7, and fixed several bugs ...
and apparently this isn't his only patch, based on the message in the email ...
Has anyone tried that? Mitani responded that he'd try it that night, but no
follow up to it, although 8 days later he answers someone else's question about
pglb hanging ...
And, is there an SVN / CVS repository for this? I checked on pgfoundry, and
the respository there hasn't been updated since '05, so obviously its moved?
Markus? Are you still out there? You are subscribed, but that doesn't mean
you don't filter things from the list to /dev/null :)
re: CyberCluster ... yes, just got word from them that they are phasing it out
due to lack of demand for it ... so that is a dead path ... :(
On 2010-02-12, at 5:43 AM, Kadi Gergely wrote:
>>> Look at the project cybercluster, they claim they have fixed these
>> bugs.
>>> And is open source too meanwhile.
>>
>> But, if its essentially pgCluster + support (which is what it looks
>> like from their web site), even if it will work on 64bit, won't it
>> suffer from the same performance / design issues that Devrim eluded to?
>
> I spent months with these tools (install, test, stress-test). It was
> working, the selects were balanced very well (our application does not have
> many writes, so it scaled very well)
>
>
> We had major problems during the recover. It crashed time-to-time, so
> unfortunately we were not able to use it in production environment.
>
> We found more bugs and reported to Cybertech, and asked for support of the
> product, but they said, they don't suggest cybercluster or pgcluster, even
> they have a programmer who worked a lot on it. What I really don't see, why
> is it still on their site as a living and supported product:( So they
> confused us as well you.
>
> One of a bugs in the balancing code: we have a table which contains a field
> "created", and it was replicated because it matches the "create" pattern.
> Mitani: you could look at it or I can send a patch for it.
>
>
>
> Regards,
> Gergo
>
> _______________________________________________
> Pgcluster-general mailing list
> [email protected]
> http://pgfoundry.org/mailman/listinfo/pgcluster-general
>
_______________________________________________
Pgcluster-general mailing list
[email protected]
http://pgfoundry.org/mailman/listinfo/pgcluster-general
