[asterisk-users] Problem with new AEX800 card dying because of interrupt problems

[Christian Weeks]

Hello
I purchased an AEX800 card to replace the ageing cheap channel bank/T1
card solution a few months ago, assuming that it would be a more robust
solution for my small scale phone system. However, it appears to be
anything but that.

Originally implemented as a XEN dom-u virtual machine on a large server
class machine, using PCI passthrough to pass the AEX800 and a small
older TDM400, then recently migrated to the dom-0, the aex800 has
continued to experience interrupt errors:

wctdm24xxp 0000:04:08.0: Missed interrupt. Increasing latency to 8 ms in
order to compensate.
wctdm24xxp 0000:04:08.0: ERROR: Unable to service card within 25 ms and
unable to further increase latency.

Eventually, it gets to be too much and the card dies:
wctdm24xxp 0000:04:08.0: Host failed to service card interrupt within
128 ms which is a hardunderun.
oh and also:
wctdm24xxp 0000:04:08.0: Power alarm on module 4, resetting!


Now, these interrupt problems have occured inside and outside a VM. So
far I've had the very unhelpful advice to "move the interrupt". Given
that this is a PCI express board, and should be delivering MSI
interrupts which are immovable, that seems to be somewhat impossible.
The BIOS certainly has nothing (and no machine I've had in the past few
years seems to have had such a feature- interrupts are programmed by the
APIC these days).

So I am asking the list, do you have any advice except perhaps to go
back to the broken channel bank? Is it really true that my modern server
class machine (quad core xeon) cannot handle the AEX800, whereas my
seven year old AMD desktop (previous host to the T1) could handle what
seems to have been about 3x the capacity? Isn't this a massive
regression?

I tried upgrading dahdi to 2.4.0 because there was promise of an
interrupt handler rewrite there for the wctdm24xxp driver, but it has
made no difference. It should also be noted that when the driver was
inside the dom-u, I got about a week's uptime from the card. In the
dom-0 I'm getting about 8 hours of uptime.

Many thanks
Christian