puppetmaster crash after upgrade to 0.24.5

[Mark D. Nagel]

On CentOS 4.6, updated puppet and puppetmaster to 0.24.5 (from dlutter
repo) and started seeing frequent process crashes with no logging. Ran
with debugging and see the following after a long series of
normal-looking lines:

debug: mount[mindseye_legacy]: Describing /mindseye_legacy/wm/lib for
mindseye.corpinfo.ms.willingminds.com
/usr/lib/ruby/1.8/yaml.rb:39: [BUG] rb_gc_mark(): unknown data type
0x0(0xbcd988) non object ruby 1.8.1 (2003-12-25) [i386-linux-gnu]

Anyone have an idea on how to deal with this? The referenced line is just:

yp = @@parser.new.load( io )

I assume this means that the YAML object is corrupted somehow, but
really not sure what else to look for to track that down yet. Or, do I
need to find a newer version of ruby? The RPM on the server is
ruby-1.8.1-7.el4_6.1. Better to run on CentOS 5 with
ruby-1.8.5-5.el5_2.3.i386.rpm? Or manually build a newer ruby RPM ala
http://sipx-wiki.calivia.com/index.php/Building_ruby-1.8.4_and_ruby-gems?

Thanks,
Mark

[Peter Meier]

Hi

> I assume this means that the YAML object is corrupted somehow, but
> really not sure what else to look for to track that down yet. Or, do I
> need to find a newer version of ruby? The RPM on the server is
> ruby-1.8.1-7.el4_6.1. Better to run on CentOS 5 with
> ruby-1.8.5-5.el5_2.3.i386.rpm? Or manually build a newer ruby RPM ala
> http://sipx-wiki.calivia.com/index.php/Building_ruby-1.8.4_and_ruby-gems?

i'm using the http://rubyworks.rubyforge.org/ rpm's without any problem.
however only on centos 5.

greets pete

[Mark D. Nagel]

Peter Meier wrote:
>> Or manually build a newer ruby RPM ala
>> http://sipx-wiki.calivia.com/index.php/Building_ruby-1.8.4_and_ruby-gems?
>>
>
>
> i'm using the http://rubyworks.rubyforge.org/ rpm's without any problem.
> however only on centos 5.

I ended up using the FC5 SRPM as described and this seemed to solve the
primary crashing problem as well as the massive memory growth on
clients. Now I am wrestling with this one spewing constantly:

Could not call fileserver.describe: #<Errno::ECONNRESET: Connection
reset by peer>

Looks like I must go to Mongrel or similar now.

Thanks,
Mark

--
Mark D. Nagel, CCIE #3177 <mna...@willingminds.com>
Principal Consultant, Willing Minds LLC (http://www.willingminds.com)
cell: 949-279-5817, desk: 714-630-4772, fax: 949-623-9854

*** Please send support requests to sup...@willingminds.com! ***