On Thu, Jul 23, 2009 at 11:33 PM, Greg<greg.b...@gmail.com> wrote:
>
> Hi all,
>
> I have an NFS server on our network sharing out package files for
> Puppet to install to all and sundry. The clients access said NFS
> server via an automounter configuration (I don't want the packages
> share mounted all the time, only when installing packages) As a
> result, installation of packages depends on autofs... So I have the
> following:
>
> service { "autofs": ensure => running", subscribe => File["/etc/
> auto_master"] }
>
> package { "SUNWfoo":
> # Yes, I'm using /net - its evil, I know...
> source => "/net/$installsvr/export/pkgs/$kernelrelease/$hardwareisa/
> SUNWfoo.pkg",
> require => Service["autofs"]
> }
>
> I'm serving it out this way for a number of reasons, including slow
> Puppet fileserving for large files, inability to specify response
> files with http package streams on Solaris' package management system
> - and so on...
>
> The problem I have is that if the autofs service is restarted right
> before a package is installed (which typically happens when I do a
> jumpstart as I put in a new /etc/auto_master) then the service is
> still restarting whilst the subsequent resources are being configured.
> So the package installs fail because they can't see the files.
>
> 5 seconds later the automounter comes online and the files are
> available...
>
> Basically the problem here is that the command used to restart the
> service doesn't wait for the service to actually restart - it just
> triggers the SMF services to restart it. I can't see any options in
> svcadm to restart and wait for it to be online again - which is a
> little annoying. And I think Puppet makes the assumption that because
> it told the service to restart that it has actually done that...
>
> If I take the requirement for autofs out from the packages - nothing
> installs as it tries to install packages too early and the system
> isn't configured enough to retrieve them...
>
> Has anyone else seen this condition happen? If so, what did you do to
> work around (or preferably fix) the problem?
>
You can specify the restart command for a service, so I would just use
your own for autofs that includes a "&& sleep 5" after the regular
restart command.
Matt
--~--~---------~--~----~------------~-------~--~----~
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+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/puppet-users?hl=en
-~----------~----~----~----~------~----~------~--~---
