[csw-maintainers] [csw-users] Munin 1.3.4 in testing

[Trygve Laugstøl]

Trygve Laugstøl wrote:
> Trygve Laugstøl wrote:
>> Juergen Arndt wrote:
>>> On Mon, 06 Jul 2009 18:32:29 +0200, Trygve Laugstøl 
>>> <trygvis at opencsw.org> wrote:
>>>
>>>> Juergen Arndt wrote:
>>>>> Hi,
>>>>>  I've put Munin 1.3.4 into testing. It's the initial version for 
>>>>> OpenCSW. The software consists of the packages munin_master and 
>>>>> munin_node.
>>>>
>>>> Also, why was 1.3.4 packaged and not the stable 1.2.6? Is 1.3.4 know 
>>>> to be as stable as 1.2.6?
>>>>
>>>> I see that Fedora and Debian both package 1.2.5/.6:
>>>>
>>>> http://dag.wieers.com/rpm/packages/munin/
>>>> http://packages.debian.org/search?keywords=munin&searchon=names&suite=all&section=all 
>>>>
>>>
>>> 1.2.6 is the last official stable version, but it's already more than 
>>> a year old and I was asked to package a more current version. If this 
>>> does not comply with our policies, I will package 1.2.6 as well, no 
>>> problem.
>>
>> No, I was just wondering really.
>>
>>>> I just tried your packages. The installation look fine and I made 
>>>> some small notes on the wiki[1]. However, when I tried to create a 
>>>> graph, all and any letters are rendered as a box. See the attached 
>>>> image. I assume that this really is an rrdtool issue more than a 
>>>> Munin issue.
>>>>  [1]: http://wiki.opencsw.org/munin-package
>>>
>>> I cannot confirm that, on my installation the graphs look as 
>>> expected. Maybe some dependencies are missing, I will try to dig into 
>>> it.
>>>
>>> I will update your wiki page in the next future with some more 
>>> installation details.
>>
>> I think it would be nice if the package included two complete apache 
>> configuration files (the basic and the one with CGI) so that the user 
>> can just Include them from httpd.conf. From what I can tell it is not 
>> likely that a user will need to adjust the configuration at all, at 
>> least not to get started.
>>
>> As for my rrd issue I just did a test from the rrd tutorial and the 
>> graphs look ok there.
> 
> I tried to use the CGI version and got a Pango related error:
> 
> [Mon Jul 06 19:21:52 2009] [error] [client 127.0.0.1] (process:6716): 
> Pango-CRITICAL **: No modules found:
> [Mon Jul 06 19:21:52 2009] [error] [client 127.0.0.1] No builtin or 
> dynamically loaded modules were found.
> [Mon Jul 06 19:21:52 2009] [error] [client 127.0.0.1] PangoFc will not 
> work correctly.
> [Mon Jul 06 19:21:52 2009] [error] [client 127.0.0.1] This probably 
> means there was an error in the creation of:
> [Mon Jul 06 19:21:52 2009] [error] [client 127.0.0.1] 
> '/opt/csw/etc/pango/pango.modules'
> [Mon Jul 06 19:21:52 2009] [error] [client 127.0.0.1] You should create 
> this file by running:
> [Mon Jul 06 19:21:52 2009] [error] [client 127.0.0.1] pango-querymodules 
>  > '/opt/csw/etc/pango/pango.modules'
> 
> Which I did and I got this output:
> 
> /opt/csw/lib/64/pango/1.6.0/modules/pango-arabic-fc.so 
> ArabicScriptEngineFc PangoEngineShape PangoRenderFc arabic:* nko:*
> [snip]
> 
> Why are there only 64-bit modules there?
> 
> Anyway, running the CGI again gave me this:
> 
> Pango-WARNING **: ld.so.1: perl: fatal: 
> /opt/csw/lib/64/pango/1.6.0/modules/pango-basic-fc.so: wrong ELF class: 
> ELFCLASS64

Ok, I've made more progress. If I ran 
/opt/csw/bin/i386/pango-querymodules I got a more sane configuration.

After that the CGI stuff started to fail, but when I went back to the 
non-CGI way it suddenly worked. Not using the CGI way is not an issue 
for me as it is not recommended at all so I'm happy.

--
Trygve