You can use the even handler system to pin pointvthe exact step thats consuning time (just inherit the base dispatcher and print current time ). From there you can use standard rbtrace or other profiler to check gc and memory/cpu related stats as well.
It may be some resource related ..like file checksum etc or ruby gc related (unlikely in your case).. but the first logical step will be to figure out the exact stage. There arr resource and recipe wise time profiler also, distributed as a chef handler
https://github.com/joemiller/chef-handler-profiler
Also, whats the output of --force-logger. That should indicate the time consuming stage as well.
Hi –We're noticing that chef solo is taking very long (9-11 seconds) to run for what seem like a small set of recipes that create a few files (we do not do any network IO as part of that chef run)Looking at the logs, it seems most of the time is spent during startup. Are there any tips and tricks to improve startup time?If it helps, we're running the latest (v11.12.8) installed via omnibus on a centos 6.5 OpenVZ guest.- Ketan
Archive powered by MHonArc 2.6.16.