Hi Brad!
(My original mail has not appeared yet on the list - I had a skew
between subscription and sender address. This is now resolved; can a
moderator approve my previous mail?)
On Thu, 19 Jun 2008, Brad Nicholes wrote:
> Please see my other post as a response to you concerns.
*nod* Even so, I still have the feeling this 3.1-release is a bit too
rushed if you have issues like the CentOS4 one remaining. For my part I
would recommend doing like Carlos suggests: Do an 'official' beta release
(with these known difficulties attached) to get a wider audience for
testing (integration into Fedora Core or even Fedora EPEL might be a good
idea). If that does not turn up anything major, make a 'real' relese of
3.1. If something comes up, you can always do a new beta release.
To keep things in perspective, I am aware that even that process will
not shake out all the bugs, but I think an effort should be made to get a
fresh set of eyes on the code.
>> I fully agree. Even if we (in Cologne) are too pressed for time and warm
>> bodies to do any testing, we would love to upgrade to 3.1. But more or
>> less becoming a beta tester is not something we would do in our production
>> environment. A stable release has to be just that: a stable release.
>
> I will say here that nobody would expect you to use your production
> environment for beta testing or even trying out a new "stable" release
> of any software. However what the Ganglia project lacks is testers. We
> have a lot of Ganglia users that rely on our tried and true 3.0.x
> version of Ganglia but we don't have many who are willing to spend time
> helping us test new features. As I have requested before on this list,
> please join us and help us make Ganglia the best it can be. You don't
> have to develop code to make a difference in the quality of the Ganglia
> software.
All of the above is true - I would really love to help. But, as I said
before, we are really tight on manpower here, so we do not have enough
resources to actually do anything (much). For example, building a new
snapshot of the source every few days simply is too much time for us.
Quite apart from the fact that we have a policy 'no make; make
install'-software. It *has* to be RPM.
If there was a set of (beta) RPMs that we could simply drop onto a few
of our client nodes we could probably do something. On the other hand, we
are still using Ganglia gmetad 2.5.7 and IIRC 3.1.x will not really work
together with this. Correct? But building a test environment is too much
work for a tool like this, unfortunately.
>> btw, I hope the python gmetad will not replace the C gmetad in the
>> future. We have seen far too many memory leaks in python applications to
>> feel comfortable with. A process that should be 10 MB in size blowing up
>> to 1.5 GB over a few weeks is not really anything we want to see again.
>
> No, there are no plans at the moment for the gmetad-python version to
> replace the existing C gmetad. And the reasons why are for exactly the
> reasons that you stated.
Good! I feel much better now.
> The purpose of the python rewrite of gmetad is
> to introduce a new architecture and plugin interface.
The plugin interface is one of the niftier ideas that I really like
about the 3.1 branch. When we finally migrate to 3.1.x, we will surely use
this interface to monitor some of the stuff we think interesting.
> If the community
> likes it and is willing to improve it, then maybe someday it will
> replace the C version. However there is a lot to be determined before
> that happens.
Well, as log as it is not a requirement for the forseeable future, I am
happy.
Thanks four your thoughts!
Dipl. Chem. Dr. Stephan Wonczak
Zentrum fuer Angewandte Informatik (ZAIK)
Regionales Rechenzentrum der Universitaet zu Koeln (RRZK)
Universitaet zu Koeln, Robert-Koch-Strasse 10, 50931 Koeln
Tel: ++49/(0)221/478-5577, Fax: ++49/(0)221/478-5590
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
Ganglia-developers mailing list
Ganglia-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/ganglia-developers
