Upgrade 2.03 to 2.1 - High CPU, multiple updaterrd.sh procs
-
First of all, thank you for this wonderful release! There are many new important features in it, we've been using pfsense since v 1.2 for our vcloud perimeter and I really love this product!
So, this is my experience upgrading from 2.03 our production carp cluster ( 6 nics, +100 vlans, 100 Mbits )
-
Last Friday, around midnight we've upgraded a production carp cluster from 2.03 to 2.1, I must say there were no problems after the reboot of the slave nor of the master apart from a slight rise in CPU usage( +20% ).
-
Saturday, around midday, CPU boosted to 100% so we've started investigating… Apparently there was a number of updaterrd.sh threads spanning multiple pfctl -ss processes that were eating the CPU ( 10 yrs old xeon server, on recent hw it would pass unnoticed );
-
On Saturday night we've rebooted both master and slave but nothing changed, we had 4 updaterrd.sh procs on master and 6 on slave
-
On Sunday morning we've stopped the rrd backed ( status -> rrd graphs -> settings ), killed all the remaining updaterrd.sh threads and restarted the rrd backend. CPU usage returned normal, as it was before the upgrade :)
I guess there are problems in the rrd part of the upgrade scripts or some sort of cron task issue.
Let’s hope it's solved, it’s Monday ;)
Cheers
Dario -