I have one node that continually generates a configuration timeout, I'm not sure why, but because it fails to generate a config in time, it times out and then reverts to the cached config. I thought I'd increase the configtimeout to something larger so that it had enough time to figure itself out, however it doesn't appear as if the option is working:
# time puppetd --configtimeout 999999 --test
info: Loading fact vserver
info: Loading fact interfaces
info: Loading fact in_colo
info: Loading fact acpi_available
info: Loading fact public_ip
info: Loading fact netmask
info: Loading fact ipmess
info: Loading fact virtual
info: Loading fact private_ip
info: Retrieving plugins
notice: Ignoring cache
err: Configuration retrieval timed out
warning: Not using cache on failed catalog
real 2m20.477s
As you can see above, the actual time that puppet sat around waiting for things before it timed out was only 2 minutes, no where close to the large amount I specified on the command-line.
Additionally, I tried setting this in puppet.conf:
[puppetd]
pluginsync=true
factsync=false
report=true
configtimeout=999999
But same behavior, approximately 2 minutes and a handful of seconds later, a configuration timeout.
I see in the master logs that the configuration compilation for this node takes around 168 seconds, which must be too long.
