FAQ
Hi,

I've built a puppet module myself to manage puppet.conf and I use a
subscribe for the service, so it restarts the client whenever puppet.conf
changes. I think I shouldn't but I'd like to have a confirmation. The
fact that foreman sees a failure upon a change in puppet.conf makes me
think even more that I am wrong...

Thanks,

--
You received this message because you are subscribed to the Google Groups "Puppet Users" group.
To view this discussion on the web visit https://groups.google.com/d/msg/puppet-users/-/13IctiK6RhQJ.
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.

Search Discussions

  • Ryan Coleman at Dec 12, 2012 at 11:13 pm

    On Wed, Dec 12, 2012 at 10:57 AM, Ugo Bellavance wrote:

    Hi,

    I've built a puppet module myself to manage puppet.conf and I use a
    subscribe for the service, so it restarts the client whenever puppet.conf
    changes. I think I shouldn't but I'd like to have a confirmation.
    The fact that foreman sees a failure upon a change in puppet.conf makes me
    think even more that I am wrong...

    If the service was successfully restarted, it should register as a
    successful change. That's normal behavior. Is that not what you see?

    Thanks,

    --
    You received this message because you are subscribed to the Google Groups
    "Puppet Users" group.
    To view this discussion on the web visit
    https://groups.google.com/d/msg/puppet-users/-/13IctiK6RhQJ.
    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.


    --
    Ryan Coleman | Modules & Forge | @ryanycoleman | ryancoleman in #puppet

    --
    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.
  • Ugo Bellavance at Dec 13, 2012 at 12:41 am

    On Wednesday, December 12, 2012 6:13:41 PM UTC-5, Ryan Coleman wrote:



    On Wed, Dec 12, 2012 at 10:57 AM, Ugo Bellavance <ug...@lubik.ca<javascript:>
    wrote:
    Hi,

    I've built a puppet module myself to manage puppet.conf and I use a
    subscribe for the service, so it restarts the client whenever puppet.conf
    changes. I think I shouldn't but I'd like to have a confirmation.
    The fact that foreman sees a failure upon a change in puppet.conf makes me
    think even more that I am wrong...

    If the service was successfully restarted, it should register as a
    successful change. That's normal behavior. Is that not what you see?

    Makes sense, but I think Foreman may have a way to determine whether the
    report is Applied, Restarted, Failed, Restart Failure, Skipped or Pending.
    Maybe it shows Failed because it doesn't see the "Finished catalog run in
    0.64 seconds" at the end? I guess maybe I should ask to the Foreman group,
    but I was mostly wondering whether it was a good idea or not to restart
    puppet when changing puppet.conf.

    Thanks,

    Ugo

    --
    You received this message because you are subscribed to the Google Groups "Puppet Users" group.
    To view this discussion on the web visit https://groups.google.com/d/msg/puppet-users/-/0HW2jDtOobsJ.
    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.
  • Matthew Burgess at Dec 13, 2012 at 10:43 am

    On Thu, Dec 13, 2012 at 12:41 AM, Ugo Bellavance wrote:

    I guess maybe I should ask to the Foreman group,
    but I was mostly wondering whether it was a good idea or not to restart
    puppet when changing puppet.conf.
    I think it would be a bad idea not to. Otherwise you've got a process
    running under a different configuration than what is on disk, so when
    it doesn't look like it's behaving like the config suggests it should,
    it can cause a lot of head-scratching!

    Note the advice in the thread at
    https://groups.google.com/d/msg/puppet-users/LViZFxc5ZAE/2yWRRixDoOQJ
    though, about how to get puppet to successfully restart the agent!

    Regards,

    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+unsubscribe@googlegroups.com.
    For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en.
  • Ugo Bellavance at Dec 14, 2012 at 12:03 am

    On Thursday, December 13, 2012 5:43:14 AM UTC-5, Matthew Burgess wrote:
    On Thu, Dec 13, 2012 at 12:41 AM, Ugo Bellavance wrote:

    I guess maybe I should ask to the Foreman group,
    but I was mostly wondering whether it was a good idea or not to restart
    puppet when changing puppet.conf.
    I think it would be a bad idea not to. Otherwise you've got a process
    running under a different configuration than what is on disk, so when
    it doesn't look like it's behaving like the config suggests it should,
    it can cause a lot of head-scratching!
    That rises two questions in my mind:


    1. In my understanding, puppetmaster catches any changes in the
    manifests without needing a restart, why can't the puppet client reload its
    puppet.conf parameters at the next run?
    2. Why puppet can't use the "reload" option of most current daemons'
    init scripts? I think there should be a parameter "hasreload", similar to
    "hasrestart", saying that the init script supports the use of the "reload"
    argument. This would avoid this kind of problem, and people wouldn't have
    to redefine the restart command.

    Ugo

    --
    You received this message because you are subscribed to the Google Groups "Puppet Users" group.
    To view this discussion on the web visit https://groups.google.com/d/msg/puppet-users/-/Vi0T1D5_sfwJ.
    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.
  • Eric Sorenson at Dec 14, 2012 at 7:08 am

    On Thursday, December 13, 2012 4:03:20 PM UTC-8, Ugo Bellavance wrote:

    That rises two questions in my mind:


    1. In my understanding, puppetmaster catches any changes in the
    manifests without needing a restart, why can't the puppet client reload its
    puppet.conf parameters at the next run?
    There's a pretty big distinction between the puppetmaster watching
    -manifests- that it uses and watching/changing -config settings- : the
    first one is pretty safe (controlled by the `filetimeout` setting and
    mostly works), the second is pretty perilous when running in a monolithic
    long-lived ruby process.

    There is a lot of tangled stuff around puppet early bootstrapping and
    initialization that is a major focus of work for the upcoming 3.1
    release: http://projects.puppetlabs.com/versions/288 but even if after
    that's cleared up, the edge cases are pretty brutal -- For example if the
    value (or heck, the contents) of $libdir changes and new versions of
    plugins are available, they can't be reliably reloaded into the same ruby
    process. See stuff like http://projects.puppetlabs.com/issues/4248#note-29
    , and the long list of related issues
    on http://projects.puppetlabs.com/issues/7316 .

    IMO this is the biggest drawback to long-running puppet daemon processes
    compared to periodic instantiation via cron / god / daemontools /etc: you
    just cannot be confident that the interpreter state is actually going to
    reinitialize cleanly if something big changes underneath it.

    1. Why puppet can't use the "reload" option of most current daemons'
    init scripts? I think there should be a parameter "hasreload", similar to
    "hasrestart", saying that the init script supports the use of the "reload"
    argument. This would avoid this kind of problem, and people wouldn't have
    to redefine the restart command.

    Look back through some of the prior discussions on this topic; I can
    understand the desire but previous discussions have raised some legitimate
    issues that nobody's code so far has addressed --

    http://projects.puppetlabs.com/issues/1014 (!!)
    http://projects.puppetlabs.com/issues/3323

    Looks like the most recent state of play was Daniel Pittman's proposal
    here: http://projects.puppetlabs.com/issues/3323#note-14
    which seems nice & clean (add support for `hasrestart => graceful`).

    eric0

    --
    You received this message because you are subscribed to the Google Groups "Puppet Users" group.
    To view this discussion on the web visit https://groups.google.com/d/msg/puppet-users/-/VJjARPE_o7sJ.
    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.

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
grouppuppet-users @
categoriespuppet
postedDec 12, '12 at 7:02p
activeDec 14, '12 at 7:08a
posts6
users4
websitepuppetlabs.com

People

Translate

site design / logo © 2022 Grokbase