Kristian Kielhofner wrote:
Jesse Guardiani wrote:
Hello,
Can anyone tell me what the "normal" number of interrupts per second is for an X100P card?
1000 / card
I've used FreeBSD 5.3 and a linux 2.6.11 kernel on the exact same hardware (only the disk changed) and `systat -vmstat 1` on FreeBSD and `procinfo -dS -n1` under Linux. For both, I'm seeing roughly 1000 interrupts per second on my X100p card. It was a bit worse under FreeBSD, and I experienced frequent lockups, hangs, and X100p malfunctions, so I switched to Linux. The machine is usable under Linux, but I still think that number of interrupts per second is a bit high.
FreeBSD had some issues with Asterisk.
This should be "has some issues". I do not consider the FreeBSD zaptel support to be production quality in any way. I experienced reproducible system hangs (mostly after an asterisk restart), interrupt issues (audio skips and SSH pauses during typing), and general instability. This was with an up-to-date FreeBSD 5.3-SECURITY and the latest zaptel at asterisk from ports (1.0.6 for asterisk, and a significantly lower version for zaptel, I think).
I do not recommend anyone run FreeBSD + Asterisk at this time.
I said FreeBSD had some issues with Asterisk, not zaptel. I have not run Zaptel on FreeBSD, so I couldn't tell you. I did, however, run Asterisk on FreeBSD 4.11 briefly and remember having to tweak modules.conf.
Also, on 2.6.11 look at the "timer" in /proc/interrupts. It's 1000/second too.
Yeah, I saw that. How is that significant? I'm a software guy, not a hardware guy, so I don't know much about interrupts. Just that 1000 interrupts/sec is fairly high. :)
My note about the kernel and 1000/sec was there to demonstrate that it is not that unusual, and that every 2.6.x system in the world (by default) has the timer doing 1000/sec.
My motherboard is an Abit BE6, and it seems to have some IRQ assignment problems, so I'm wondering what my baseline should be.
Also, I have a dual CPU PII motherboard with two X100P cards in it, and it's hitting about 1000 interrupts per second per card too. Is this normal? How many interrupts per second can a given CPU sustain?
If you want to reduce interrupt load, go down to one card. If you have 4 X100P's, that 4000 interrupts/sec. If you have 1 TDM400, that's 1000/sec.
Again, is there a formula to describe how many interrupts per second a given CPU, PCI bus, and FSB bus can theoretically sustain per second?
I'm curious if my 450mhz PIII should be able to handle 2 x100p cards, or just 1? What's the limit? Etc....
I don't know of a formula, but look at the output of "System" in top to find out. If you have too many interrupts, it will be very high and the system will be very, very sluggish. Other than that, I don't know of any way to tell.
But my note before was suggesting no more than one Digium card / machine, because it is messy to be generating all of those interrupts when Asterisk just needs one source for timing.
But yes, at this time I see no reason to run * on FreeBSD. As I have said before * can be tough as it is, don't through a less common OS (for Asterisk) in the mix. Your capabilites are limited, there are fewer people to help you, etc. Did you try any of this on the -bsd list?
-- Kristian Kielhofner _______________________________________________ Asterisk-Users mailing list Asterisk-Users@lists.digium.com http://lists.digium.com/mailman/listinfo/asterisk-users To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users
