Uploaded image for project: 'Puppet'
  1. Puppet
  2. PUP-9563

Using max_queued_requests, agents return to originally scheduled run times



    • Bug
    • Status: Ready for Engineering
    • Normal
    • Resolution: Unresolved
    • PUP 5.5.17, PUP 6.11.1
    • None
    • None
    • Needs Priority
    • 33803,47856
    • 2
    • Needs Assessment


      Puppet Enterprise 2018.1.3

      Puppet Version: 5.5.3
      Puppet Server Version: 5.3.4**
      OS Name/Version: Rhel 6.9

      Raised on behalf of a PE customer, on using max_queued_requests the customer has noticed that after an agent receives the 503 error, it then waits the random amount of time and does its puppet run. The agent then returns to its regular scheduled agent runs times.

      Describe steps to reproduce…

      Reproduced this locally, I had a 2018 master, I spun up 3 agents, reduced jruby_puppet_max_queued_requests to 1 and jruby_max_active_instances to 1. I then ran the agent simultaneously on the 3 agents causing 1 of them to hit the 503 error. After this agent waited the random amount of time before doing a puppet run, however, it returned to its original 30-minute interval run after this:

      Regular daemonized runs:

      503 and run after wait time:

      Agent returns to daemonized run time:

      Desired Behavior: When the max_queued_requests limit is hit and agents delay their run time, after this, their regular scheduled daemonized run should run at 30-minute intervals from the last run, not the original run times.

      Actual Behavior: After the 503 is received, agent run times return to the regular scheduled daemonized run tmes.



        Issue Links



              Unassigned Unassigned
              patrick.grant Patrick Grant
              1 Vote for this issue
              16 Start watching this issue



                Zendesk Support