High CPU Utilization After Upgrade 2.4-RC to 2.4.2-p1

dbosiljevac

Wondering if anyone else has seen this behaviour…

Yesterday, I upgraded an SG-1000 from 2.4-RC to 2.4.2-p1. Ever since then I've been noticing high CPU utilization. I have posted a graph (last 48 hours) from one of my monitoring systems that shows the increase in CPU activity.

Not 100% sure, but System Activity seems to show php-fpm at the top of the list.

Anyone else seeing this? Any thoughts? I have restarted PHP-FPM as well as restarted the actual firewall itself, but it always comes back with this high CPU.

Thanks,

Dave
![Screen Shot 2017-12-20 at 09.16.54 AM.png](/public/imported_attachments/1/Screen Shot 2017-12-20 at 09.16.54 AM.png)
![Screen Shot 2017-12-20 at 09.16.54 AM.png_thumb](/public/imported_attachments/1/Screen Shot 2017-12-20 at 09.16.54 AM.png_thumb)
![Screen Shot 2017-12-20 at 09.17.49 AM.png](/public/imported_attachments/1/Screen Shot 2017-12-20 at 09.17.49 AM.png)
![Screen Shot 2017-12-20 at 09.17.49 AM.png_thumb](/public/imported_attachments/1/Screen Shot 2017-12-20 at 09.17.49 AM.png_thumb)

jimp

That particular output doesn't show much. A few percent used by PHP, most likely because you're viewing that from the GUI.

Connect to the console or ssh, start a shell and run "top -aSH" and wait a bit and see what shows up there.

dbosiljevac

Looks like my issue was being logged into the web portal. When I logged out of the web portal, my CPU utilization went down to average around 10%.

jimp

That is probably due to a netstat bug that was addressed a week or so ago. It will be better in 2.4.3, you can try a snapshot now if you like.

fibrewire

With pfSense 2.4.2-p1 on a Watchguard XTM 525 with celeron e3400 @ 2.6GHz i'm seeing 99% cpu usage after a few days. I've had to reboot a few times these last few weeks since the release. Is there anything I can provide to help identify this issue?

fibrewire

Looks like I may have found the issue. Under System > Advanced > Miscellaneous > RAM Disk Settings - I unchecked the "Use memory filesystem for /tmp and /var" and the issue seems to have gone away, with the CPU idling around 17% - but only time will tell.