so we finally found a quiet night to migrate our production environment
from chef server 10 to chef server 11.
we were previously running 10.16.2 on a dual core 8gb virtual machine,
serving about 450 VMs that each checked in every 15 minutes (with start
times randomized across the 15 minutes).
Over a 4 hour window, the VM would be severely stressed and we would
normally see between 2000 and 3000 timeouts and/or connection resets trying
to connect to the chef server, and chef runs that normally took around 25
seconds would take 5 minutes.
Over the same 4 hour window last night, there were zero timeouts, zero
connection resets, and the new chef 11 VM (same specs) was barely breaking
a sweat.
Oh yeah, and the average run time over all 450 vms went from 230 seconds to
17 seconds.
I would call that a serious win and vote of confidence in the
erlang/postgresql rewrite.
Thanks guys for some amazing work!
-Jesse