CPU Reaches 100% every hour

pburgisser

Hi everyone !

I have a strange behavior on my PFSense (OpenAlliance Appliance) : every hour during ~5 min, the CPU reaches 100%. So PFSense doesn't deserve any requests (particularly the DNS forwarder).

Using SSH, I did a "ps -aux" and I could see :

USER PID %CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
root 25 73.0 0.0 0 8 ?? WL 8:10AM 124:24.91 [irq15: vr2 ata1]
root 10 15.0 0.0 0 8 ?? RL 8:10AM 1223:53.09 [idle]
root 23 11.0 0.0 0 8 ?? WL 8:10AM 28:02.36 [irq10: vr0]

This appears even if I reboot the FW.
I don't have any services such as squid, vpn or other heavy services running

Any idea?

Thank you for your help

stephenw10

What version of pfSense are you running; 1.2.3? 2.0Beta? Embedded? Nano? Full install? Live CD?

Steve

pburgisser

1.2.3-RELEASE
built on Sun Dec 6 23:21:36 EST 2009

stephenw10

Is that Nanobsd or full install?

Has it always behaved like this or just started doing it?

pburgisser

Hmmm, It's a good question, I guess it's a full install.
I bought the appliance with PFSense included on:
http://www.applianceshop.eu/index.php/firewalls/opnsense/rackmountable-opnsense/opnsense-pfsense-19-appliance.html

Thanks for your help

stephenw10

Ah well you shouldn't be having any trouble with that! :)

If it's NanoBSD it will say so on the 'system overview' page, when you first log in to the web GUI, under 'platform'.

I'm not sure I can help you too much since all my experience is with NanoBSD but answering these questions will no doubt help someone else to help you!

Have you just bought it, has it always done this?

Steve

pburgisser

it's written "pfsense" under platform :-/
No, it used to work perfectly and it has been going down since 2 or 3 months…

stephenw10

I think it says that for a full install.
What sort of hard disk is in this box?

jimp

Looks like one of your NICs (vr2) and a hard drive controller (ata1) are sharing an IRQ, and one of them is generating a lot of interrupts.

It could be a lot of network activity on that interface, it could be a lot of hard drive activity on that controller, or it could be a driver bug due to them being shared.

If possible, check your system logs/dmesg output to see what is on ata1, and look in the BIOS to see if there are any options for the hdd controller. You might be able to nudge the devices onto different IRQ's, or something similar. If you have nothing attached to ata1, you might even be able to disable the secondary ATA controller in the BIOS.

wallabybob

@jimp:

or it could be a driver bug due to them being shared.

Its a bad idea to share a level triggered interrupt (PCI device) with an edge triggered interrupt (ata and ISA devices). vr0 should be moved to its own IRQ or one shared with other PCI devices (USB, vr1, etc etc). There are often mechanisms in the BIOS for adjusting interrupt lines but they are usually very system specific. For example, you might have to specify something like "IRQ15 is not available to PCI".