Performance issues since Update 2.2.1

  • Hi!

    since the last pfSense Update i noticed some quite strang behaviour in my pfSense: For no reason (besides the pfSense update nothing was changed, pfSense is running within ESXi) i am having high latency between my VM and the pfSense! Normally (which menas before the Update) these values should be around 1ms (rather less) - no from time to time i have latencies over 100ms! It's sometimes like 20 good packets (~1ms) and than 2-10 bad packets (70ms or more)…

    Here is the thing: My pfSense has different LAN-Interfaces (6x). Behind each Interface is 1x virtual machine. The problems do not occur simultaneously on each interface at the same time, it's like the effect is jumping around from interface to interface... the only logic i got so far is that disabling and re-enable the interface will reduce the issue for some time before it's back again... in general neither the ESXi nor the pfSense it self has high load.

    Second strang thing i noticed: Related to each LAN-Interface is a VPN (IPSec) connection in order to connect some external phones (SIP) to the VM (PBX). Yesterday i experienced another strange issue: From like 8 connected and working phones only one Phone had no connection to the PBX so i checked my Firewall: Nothing was dropped! After some research i disabled the Firewall scrub - Phone worked again. Today i had another Phone (Snom, working via STUN, behind NAT) which was not working anymore since i disabled the Firewall scrub option- in this case i was able to see that my Firewall was blocking my INVITE-packets for no reason. On the same network is another phone working exactly the same way as the one beeing blocked. Except the fact that this one is not beeing blocked - why? I am 100% sure there is no differnce in theire configuration... Reversing the Firewall scrub option will make the phone work again - which off course will cause the first phone to not work anymore...

    As soon as i restore the old pfSense version all the issues are gone, so i assume this is a bug. Does anyone else have similar issues?

Log in to reply