Puppet users - Feb 2009 - Am I the only one with lots of problems with puppet?

Hi all,

I''m facing many problems for last days with my puppet server/client.
Those who read mail list may know that.


I''m not sure if it''s just my and my poor knowledge on puppet
(I must
say in my defense that I have read many wiki docs and I have bough the
book, ok, it''s not enough, if I''m a bad sysadmin I can read a
complete
library and know anything...) but seems that every doc describes one
behave and puppet does an other. Please, don''t feel offended
after/before reading this mail. I really appreciate the developers
effort, and all the help I got from this list, but I''m getting crazy
with puppet.


I started with a simple scenario, a couple of nodes with a server, and
few classes, all worked fine... But one day I faced first problem:

err: Could not retrieve catalog: Could not parse for environment production

after reading a lot, checking my conf one time and other, and other,
and other I realized that I removed puppetmasterd link
from /etc/puppet/ (puppetmaster complained about having two files with
same conf) and the solution was creating it again, ignoring the log
error...

After that, I upgraded to a newer version, not latest, where I found a
installation bug about creating $vardir directory... I could solve it
by creating it by hand, but I faced some other problems like that
version was not able to run puppetmasterd in debug mode and not sending
the process to background... So, I have to upgrade again, cause I think
this feature is basic for debugging... OH! and I found that each
section of puppetd needs its own variables... $vardir was default value
for puppetmasterd if I did not specify them explicity.

So, now I''m at version 0.24.7, from epel, and all seemed to work fine
again (complete puppetview report, many classses with deps, etc...)

until when? until I tried to configure mongrel (ok, it''s mongrel, not
puppet itself), but balancing did not work, at least in my case
(it did balance, but was not able to server catalog to hosts). I tried
to reinstall 40 host at once and I saw puppetmasterd being not able
to support that... 100 apache processes...) and it''s only 40 nodes (ok,
someone is managing lots of nodes, someone told me that they have more
than 100), my question is: did you do it with "regular" versions? did
you find many problems/bugs? did you do at once? 
I thought, balancing is not working, maybe is a proxy/mongrel
missconfig, I''ll go to mongrel list and ask them... but when I ran
puppetca w puppet/mongrel I cannot clean certs, so puppetca does not
work, I must locate certs and remove them by hand...

So I had to reisntall nodes from 10 to 10, and worked fine... until
when? until today, yes, I faced an old known problem:

Could not retrieve catalog: Could not parse for environment production: No
file(s) found for import of

*this time the link did not solve the problem...

Ok, you''ll say: "you have changed many things..." but no,
I haven''t! I have all my code under subversion, I go back to a
"valid" (yesterday''s) revision, I check my changes and I have
only
added a couple of packages types... but, anyway, I go back there, and
what happens now?

[root@gridinstall tmp]# /usr/bin/ruby /usr/sbin/puppetmasterd
--manifest=/etc/puppet/manifests/site.pp --no-daemonize --debug
[root@gridinstall tmp]# !ps
ps -ef|grep puppet
puppet   14551     1  0 19:05 ?        00:00:00 /usr/bin/ruby
/usr/sbin/puppetmasterd --manifest=/etc/puppet/manifests/site.pp --no-daemonize
--debug
root     14555  2750  0 19:06 pts/0    00:00:00 grep puppet

--no-daeemonize is not working again!!!

So, do you understand what am I felling? I mean, it''s only me? am I
alone eith many problems?
I''ve checked reductivelabs prizes for support & implementation
services, maybe this is the solution, maybe not... I''ll ask my boss for
money to attend belgium''s training course... I don''t know what
else
do... but now I feel that I was using a good & easy software that I was
able to learn and deploy in my work place fast, but when I tried to do
some complex conf, I faced so many problems that I''m afraid of its
behaviour...


Anyway, again, thanks to developer and people who helped me in list,
and thanks to those who are still reading this post :-) 
I just wanted to share my recent experience with puppet... I hope in
future all my problems will disappear.

Cheers,
Arnau 


--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Puppet Users" group.
To post to this group, send email to puppet-users@googlegroups.com
To unsubscribe from this group, send email to
puppet-users+unsubscribe@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/puppet-users?hl=en
-~----------~----~----~----~------~----~------~--~---

On Wed, Feb 11, 2009 at 07:25:54PM +0100, Arnau Bria
wrote:
> I started with a simple scenario, a couple of nodes with a server, and
> few classes, all worked fine... But one day I faced first problem:
> 
> err: Could not retrieve catalog: Could not parse for environment
> production
> 
> after reading a lot, checking my conf one time and other, and
> other, and other I realized that I removed puppetmasterd link from
> /etc/puppet/ (puppetmaster complained about having two files with
> same conf) and the solution was creating it again, ignoring the
> log error...
OK, you''ve got something really strange going on.  I have noting in
/etc/puppet named puppetmasterd or anything like it.  I have a
section of /etc/puppet/puppet.conf named puppetmasterd, though.
Could you perhaps tar up your entire /etc/puppet/ and post it
somewhere?
> After that, I upgraded to a newer version, not latest, where I found a
> installation bug about creating $vardir directory... I could solve it
> by creating it by hand, but I faced some other problems like that
> version was not able to run puppetmasterd in debug mode and not sending
> the process to background... 
You had a version of puppetmasterd in which
--no-daemonize --debug didn''t work?  Which version would that be?
> So, I have to upgrade again, cause I think this feature is basic
> for debugging... OH! and I found that each section of puppetd
> needs its own variables... $vardir was default value for
> puppetmasterd if I did not specify them explicity.
I have *no* idea what you''re talking about, nor do I know what
$vardir means here.
> until when? until I tried to configure mongrel (ok, it''s mongrel,
not
> puppet itself), 
I haven''t used mongrel with puppet specifically, but every
experience I''ve had with it has been horrible.
> balancing did not work, 
If you need load balancing, look at haproxy
> So I had to reisntall nodes from 10 to 10, and worked fine... until
> when? until today, yes, I faced an old known problem:
> 
> Could not retrieve catalog: Could not parse for environment production: No
file(s) found for import of
No files found for import of what?  Looks like you''re missing the
end of the line there.  Again, can you post the whole tarball?
> [root@gridinstall tmp]# /usr/bin/ruby /usr/sbin/puppetmasterd
--manifest=/etc/puppet/manifests/site.pp --no-daemonize --debug
> [root@gridinstall tmp]# !ps
> ps -ef|grep puppet
> puppet   14551     1  0 19:05 ?        00:00:00 /usr/bin/ruby
/usr/sbin/puppetmasterd --manifest=/etc/puppet/manifests/site.pp --no-daemonize
--debug
> root     14555  2750  0 19:06 pts/0    00:00:00 grep puppet
> 
> --no-daeemonize is not working again!!!
That''s ... impressive.  Does /etc/puppet/puppet.conf specify
daemonize, perhaps?
> So, do you understand what am I felling? I mean, it''s only me? 
I have not had any of the problems you''re describing, but then
I''ve
never had any reason to run it with --no-daemonize or use mongrel.

-Robin

-- 
They say:  "The first AIs will be built by the military as weapons."
And I''m thinking:  "Does it even occur to you to try for something
other than the default outcome?" -- http://shorl.com/tydruhedufogre
http://www.digitalkingdom.org/~rlpowell/ *** http://www.lojban.org/

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Puppet Users" group.
To post to this group, send email to puppet-users@googlegroups.com
To unsubscribe from this group, send email to
puppet-users+unsubscribe@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/puppet-users?hl=en
-~----------~----~----~----~------~----~------~--~---

Arnau,

Please don''t be offended by my response, as you requested we not be
offended by your post.

I''ve watched a lot of your posts to the list, and the behavior
I''m
observing is when you run into errors, you don''t seem to step back and
try to take the time to understand them. Instead, it appears you do
the following:

1) Immediately post a message to the list.
2) Freak out because nobody responded within 5 minutes.
3) Flail around trying things, seemingly at random, until something
*seems* to work because it makes the specific error change/go away.
4) Repeat.

So, if I were to offer you some ideas for how to proceed, they would be:

1) Set all the stuff you''ve made to this point aside. Keep it around
to use as reference.
3) Pick 3 machines to use as an environment for learning puppet.
4) Ensure that Puppet is *completely* removed from these machines. It
would be optimal if they were totally clean machines.
 * Puppet is still pre-1.0 software. The upgrade path hasn''t been
perfectly clean and, to complicate things, packaging has been fairly
inconsistent. There are some artifacts that are in older versions
which will utterly break newer versions in disorienting ways.
5) Install the latest stable version of Puppet (0.24.7 as I write).
Choose one machine to act as puppetmaster, and (if it is a separate
package for your environment) install puppetmaster on that machine.
6) Pick one service. Create a module for it.
7) When you encounter a problem, stop. Walk away entirely for ten
minutes. Come back to it and see if you can figure out what the root
cause of the problem is. Some basic troubleshooting steps for Puppet:
 * Re-start your puppetmaster and watch the logs. Often a syntax
error in your manifests will cause the puppetmaster to ignore them and
use the old cached configurations. When you restart, if you are
watching the logs, you''ll catch the syntax error and you can fix it.
 * Utilize --debug --no-daemonize on both client and server -- this
is one of your biggest complaints, and it is entirely caused by having
used old versions and leaving old broken config files around. You
should not have a ''puppetd.conf'' or a
''puppetmasterd.conf'', you should
be using ''puppet.conf''
 * Try out --parse-only
 * Comment out the sections of your manifest that you added since
things ran perfectly. Uncomment one resource at a time.

I think if you step back and start from a reasonably clean slate you
will have a more positive experience.

Regards,
Paul Lathrop

On Wed, Feb 11, 2009 at 10:25 AM, Arnau Bria <arnaubria@pic.es>
wrote:
>
> Hi all,
>
> I''m facing many problems for last days with my puppet
server/client.
> Those who read mail list may know that.
>
>
> I''m not sure if it''s just my and my poor knowledge on
puppet (I must
> say in my defense that I have read many wiki docs and I have bough the
> book, ok, it''s not enough, if I''m a bad sysadmin I can
read a complete
> library and know anything...) but seems that every doc describes one
> behave and puppet does an other. Please, don''t feel offended
> after/before reading this mail. I really appreciate the developers
> effort, and all the help I got from this list, but I''m getting
crazy
> with puppet.
>
>
> I started with a simple scenario, a couple of nodes with a server, and
> few classes, all worked fine... But one day I faced first problem:
>
> err: Could not retrieve catalog: Could not parse for environment production
>
> after reading a lot, checking my conf one time and other, and other,
> and other I realized that I removed puppetmasterd link
> from /etc/puppet/ (puppetmaster complained about having two files with
> same conf) and the solution was creating it again, ignoring the log
> error...
>
> After that, I upgraded to a newer version, not latest, where I found a
> installation bug about creating $vardir directory... I could solve it
> by creating it by hand, but I faced some other problems like that
> version was not able to run puppetmasterd in debug mode and not sending
> the process to background... So, I have to upgrade again, cause I think
> this feature is basic for debugging... OH! and I found that each
> section of puppetd needs its own variables... $vardir was default value
> for puppetmasterd if I did not specify them explicity.
>
> So, now I''m at version 0.24.7, from epel, and all seemed to work
fine
> again (complete puppetview report, many classses with deps, etc...)
>
> until when? until I tried to configure mongrel (ok, it''s mongrel,
not
> puppet itself), but balancing did not work, at least in my case
> (it did balance, but was not able to server catalog to hosts). I tried
> to reinstall 40 host at once and I saw puppetmasterd being not able
> to support that... 100 apache processes...) and it''s only 40 nodes
(ok,
> someone is managing lots of nodes, someone told me that they have more
> than 100), my question is: did you do it with "regular" versions?
did
> you find many problems/bugs? did you do at once?
> I thought, balancing is not working, maybe is a proxy/mongrel
> missconfig, I''ll go to mongrel list and ask them... but when I ran
> puppetca w puppet/mongrel I cannot clean certs, so puppetca does not
> work, I must locate certs and remove them by hand...
>
> So I had to reisntall nodes from 10 to 10, and worked fine... until
> when? until today, yes, I faced an old known problem:
>
> Could not retrieve catalog: Could not parse for environment production: No
file(s) found for import of
>
> *this time the link did not solve the problem...
>
> Ok, you''ll say: "you have changed many things..." but
no,
> I haven''t! I have all my code under subversion, I go back to a
> "valid" (yesterday''s) revision, I check my changes and I
have only
> added a couple of packages types... but, anyway, I go back there, and
> what happens now?
>
> [root@gridinstall tmp]# /usr/bin/ruby /usr/sbin/puppetmasterd
--manifest=/etc/puppet/manifests/site.pp --no-daemonize --debug
> [root@gridinstall tmp]# !ps
> ps -ef|grep puppet
> puppet   14551     1  0 19:05 ?        00:00:00 /usr/bin/ruby
/usr/sbin/puppetmasterd --manifest=/etc/puppet/manifests/site.pp --no-daemonize
--debug
> root     14555  2750  0 19:06 pts/0    00:00:00 grep puppet
>
> --no-daeemonize is not working again!!!
>
> So, do you understand what am I felling? I mean, it''s only me? am
I
> alone eith many problems?
> I''ve checked reductivelabs prizes for support & implementation
> services, maybe this is the solution, maybe not... I''ll ask my
boss for
> money to attend belgium''s training course... I don''t know
what else
> do... but now I feel that I was using a good & easy software that I was
> able to learn and deploy in my work place fast, but when I tried to do
> some complex conf, I faced so many problems that I''m afraid of its
> behaviour...
>
>
> Anyway, again, thanks to developer and people who helped me in list,
> and thanks to those who are still reading this post :-)
> I just wanted to share my recent experience with puppet... I hope in
> future all my problems will disappear.
>
> Cheers,
> Arnau
>
>
> >
>
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Puppet Users" group.
To post to this group, send email to puppet-users@googlegroups.com
To unsubscribe from this group, send email to
puppet-users+unsubscribe@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/puppet-users?hl=en
-~----------~----~----~----~------~----~------~--~---

On Wed, 11 Feb 2009 13:40:56 -0800
Paul Lathrop wrote:
> 
> Arnau,
Hi Paul,
 
> Please don''t be offended by my response, as you requested we not
be
> offended by your post.
I won''t, don''t worry.  I hope you or other developers are not
angry
with me... I''m just explaining my experience with this software. But,
as I will comment later, I''ve been really happy with it till now, that
I''m a little disappointed (but I''ll continue working cause I
think it
has to work fine, many people in list say so).
 
> I''ve watched a lot of your posts to the list, and the behavior
I''m
> observing is when you run into errors, you don''t seem to step back
and
> try to take the time to understand them. Instead, it appears you do
> the following:
> 
> 1) Immediately post a message to the list.
> 2) Freak out because nobody responded within 5 minutes.
> 3) Flail around trying things, seemingly at random, until something
> *seems* to work because it makes the specific error change/go away.
> 4) Repeat.
Yep, could be, in part ;-). 
I can ensure you that I''ve wasted a lot of time trying to understand
confs and trying to understand problems. 
But, I accept that I go to list too quick,  and I post many things.
(I''m used to, cause maybe someone sees something that I don''t,
and a little comment gives me the answer... )
Random? don''t believe it, I always look for docs or previous post in
list... As I said before, _seems_ that what I find in docs and the
behave I get is not always the same. 

But Paul, I accept that my behave could be paranoid sometimes...

> So, if I were to offer you some ideas for how to proceed, they would
> be:
> 
> 1) Set all the stuff you''ve made to this point aside. Keep it
around
> to use as reference.
> 3) Pick 3 machines to use as an environment for learning puppet.
> 4) Ensure that Puppet is *completely* removed from these machines. It
> would be optimal if they were totally clean machines.
>  * Puppet is still pre-1.0 software. The upgrade path hasn''t been
> perfectly clean and, to complicate things, packaging has been fairly
> inconsistent. There are some artifacts that are in older versions
> which will utterly break newer versions in disorienting ways.
> 5) Install the latest stable version of Puppet (0.24.7 as I write).
> Choose one machine to act as puppetmaster, and (if it is a separate
> package for your environment) install puppetmaster on that machine.
> 6) Pick one service. Create a module for it.
> 7) When you encounter a problem, stop. Walk away entirely for ten
> minutes. Come back to it and see if you can figure out what the root
> cause of the problem is. Some basic troubleshooting steps for Puppet:
>  * Re-start your puppetmaster and watch the logs. Often a syntax
> error in your manifests will cause the puppetmaster to ignore them and
> use the old cached configurations. When you restart, if you are
> watching the logs, you''ll catch the syntax error and you can fix
it.
>  * Utilize --debug --no-daemonize on both client and server -- this
> is one of your biggest complaints, and it is entirely caused by having
> used old versions and leaving old broken config files around. You
> should not have a ''puppetd.conf'' or a
''puppetmasterd.conf'', you should
> be using ''puppet.conf''
>  * Try out --parse-only
>  * Comment out the sections of your manifest that you added since
> things ran perfectly. Uncomment one resource at a time.
> 
> I think if you step back and start from a reasonably clean slate you
> will have a more positive experience.
Paul, that''s exactly what I''ve done. I can comment point by
point all
your advices (or I could show you my svn logs). I''ve been working with
puppet for two or three months, building module by module until I got a
great install -> config method. It worked really fine and I was really
on love with puppet: easy and clean system.

Maybe you won''t believe me, but I have had 10 hosts running under
puppet
perfectly for more than a month ! And I''ve been really happy
with puppet! So I believe in it. 

but how? with no mongrel, with separate install time, with
no need to use a big debug (as you say, running debug on clients
always point out syntax errors, I also have pre-commit hook that
does the job for me), with no updates (i started the pre-prod
with 0.24.7 version), and mainly, with few nodes... (as you can imagine,
I cannot reinstall 40 nodes every day, so I cannot face big problems
every day). I found many problems at one time, many surprise for
what I was preparing for real production... 

But no problem, I''ll follow them again, I''ll look for
"the error" going
from svn revision to revision (I started with web''s example :-)), and
I''ll com back with the error (I suppose you would like to see it :-) ).

Paul, I really appreciate you reply, very polite and instructive. And
again, sorry if you or some other dev feel offended, it''s not my
intention, and maybe my English is not as good as I''d like. 
 
> Regards,
> Paul Lathrop
Cheers,
Arnau

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"Puppet Users" group.
To post to this group, send email to puppet-users@googlegroups.com
To unsubscribe from this group, send email to
puppet-users+unsubscribe@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/puppet-users?hl=en
-~----------~----~----~----~------~----~------~--~---