Create a Post
cancel
Showing results for 
Search instead for 
Did you mean: 
Dyslexic155
Participant

CoreXL connections not distributed correctly between the cores

Jump to solution

Hello Experts,

Yesterday we had an issue where the connections were not distributed correctly by the CoreXL.

High CPU was noticed in few cores, and after checking fw ctl multik we saw that coreXL was distributing only the connections to a few cores and not to all.

The issue was solved by rebooting the firewall.

Is it possible to know the reason of the issue and can it happen again?

I have attached output for fw ctl multik before the reboot, and more commands after the reboot.

We are running R80.30

Thanks in advance,

 

0 Kudos
1 Solution

Accepted Solutions
Timothy_Hall
Champion
Champion

This is a known issue fixed in R80.30 Jumbo HFA Take 219+, a workaround is also available as described here:

sk168513: CPU consuming on some instances is high after installing R80.30 Jumbo Hotfix Accumulator T...

 

"Max Capture: Know Your Packets" Video Series
now available at http://www.maxpowerfirewalls.com

View solution in original post

2 Replies
Timothy_Hall
Champion
Champion

This is a known issue fixed in R80.30 Jumbo HFA Take 219+, a workaround is also available as described here:

sk168513: CPU consuming on some instances is high after installing R80.30 Jumbo Hotfix Accumulator T...

 

"Max Capture: Know Your Packets" Video Series
now available at http://www.maxpowerfirewalls.com

View solution in original post

genisis__
Advisor

Tim,

As there does not appear to be a way to use SNMP to monitor a specific VS CPU usage (not that I can see, easily using enterprise standard tools) , I was wondering how we can actually confirm how many CPUs to actually allocated a VS?

Using 'fw ctl multik stat', looking at the peaks connections are there any guidelines to determine recommended peaks/connections per core?

Lets say I've allocated 10 cores to a VS and top is reporting 600% utilisation, does this mean I can safely reduce the cores down to 8 (leaving 2 cores for spikes)?

Without history graphing is seems almost impossible to balance the right number of cores and forward plan core requirements.

 

0 Kudos