https://www.puppet.com/community logo
Join Slack
Powered by
# voxpupuli
  • v
    nods
  • v
    so shall it be!
  • v
    https://github.com/voxpupuli/puppet-jenkins/pull/1065 I kind of expect it to fail, given the plugins are just a pain to maintain but I just need the puppet/systemd version limit bump
  • y
    augeas modules need some rubocop checks too.. my VSCodium complains a lot
  • y
    so my vote is for voxpupuli-rubocop gem
  • y
    (or maybe voxpupuli-rubocop-config)
  • v
    I think its fine to release it anyways
  • y
    yes, but I’d add a WARNING there
  • y
    let me confirm it’s really broken on P7
  • y
    we can release it for P6 safely though (with limiting Puppet version to <7)
  • y
    aha.. so there is no acceptance test for multiple sysctl{} changing the same /etc/sysctl.conf.. let me add one to check it’s broken there as well 😄
  • y
    forgot about articles in English 😄
  • y
    hmm.. it works fine for /etc/sysctl.conf 🤔 maybe it’s broken just for non-existent file? checking..
  • v
    a warning sounds like a good thing
  • y
    yes, it works fine when file exists before applying it 1st time
  • y
    ewoud: any idea what might be broken wrt this new info?
  • v
    oh, that's interesting - so the problem is that /etc/sysctl.conf doesn't exist before the run starts and only when it does it's created?
  • y
    Copy code
    file { '/etc/sysctl.d/20-fs.conf':
                replace  => false,
                content => '',
              }

              sysctl { 'fs.nr_open':
                value  => '100001',
                target => '/etc/sysctl.d/20-fs.conf'
              }

              sysctl { 'fs.inotify.max_user_watches':
                value  => '8193',
                target => '/etc/sysctl.d/20-fs.conf'
              }
  • y
    this works fine
  • y
    remove 
    file
    and it breaks
  • y
    I’ll add to PR too
  • v
    you don't need to manage content, just ensure => file is sufficient
  • y
    I was too lazy to read the file resource doc 😄
  • y
    I was too lazy to read the file resource doc 😄
  • v
    could it be that augeas can't/doesn't create the file if needed?
  • y
    I’d expect race condition again.. so it creates file on 1st sysctl change but overrides it with next one maybe.. (just guessing)
  • y
    I’d craft some corner cases but it’s going to be time consuming..
  • y
    it calls augeas binary internally, right?
  • v
    I think it uses the ruby library, so probably not the binary
  • y
    ah.. maybe.. it was looking for headers IIRC
1...406407408...648Latest