When attempting a "run puppet" from the UI (1.11.0 - thanks for the improved UI speed, btw) on a hostgroup (same config/installs/OS, etc), the result will be a "Failed to apply catalog: Broken pipe - " error. All hosts are Ubuntu Trusty. Here's a syslog output from a UI Puppet Run on 20 nodes:
Apr 14 11:34:27 pn02 puppet-agent[45865]: Retrieving pluginfacts
Apr 14 11:34:27 pn02 puppet-agent[45865]: Retrieving plugin
Apr 14 11:34:28 pn02 puppet-agent[45865]: Loading facts
Apr 14 11:35:15 pn02 puppet-agent[45865]: Caching catalog for pn02.blahblah.org
Apr 14 11:35:22 pn02 puppet-agent[45865]: Failed to apply catalog: Broken pipe - <STDOUT>
However, when running puppet manually from the node, or when the regularly scheduled puppet run executes, the run will complete successfully. Additionally, individual Puppet Runs, as well as runs on, e.g. 5 nodes from the UI also succeed - it's just the attempt to execute on 20 nodes that produces errors.
Any thoughts on next steps to diagnose? Is this network-congestion related? Should my master's hardware be boosted?
I was having the same issue! I was using puppetssh method to run the
puppet agent -t
command.The error would only show up when running on many hosts.
I was able to fix this by redirecting the STDOUT to /dev/null. I changed my 'puppetssh_command:' to
sudo puppet agent -t &>>/dev/null
.