[asterisk-ss7] MTP2 CRC errors resulting in flapping link

Patrick asterisk-list at puzzled.xs4all.nl
Mon Jul 3 07:59:07 MST 2006


On Mon, 2006-07-03 at 15:33 +0200, Kai Militzer wrote:
[snip]
> And here the interrupts:
> 
> ss7-gate1:/var/log/asterisk# cat /proc/interrupts
>             CPU0       CPU1
>    0:  986934556          0    IO-APIC-edge  timer
>    1:       1649          0    IO-APIC-edge  i8042
>    7:          2          0    IO-APIC-edge  parport0
>    9:          0          0   IO-APIC-level  acpi
>   14:  182607673        347    IO-APIC-edge  ide0
>   50:          0          0         PCI-MSI  pciehp
> 177:          0          0   IO-APIC-level  uhci_hcd
> 185:          0          0   IO-APIC-level  uhci_hcd, ehci_hcd
> 193:  986078651          0   IO-APIC-level  uhci_hcd, wct2xxp
> 201:          0          0   IO-APIC-level  uhci_hcd
> 209:    2640734          0   IO-APIC-level  eth1
> 225:   23208735          0   IO-APIC-level  eth0
> NMI:          0          0
> LOC:  986900234  986900230
> ERR:          0
> MIS:          0

Digium cards are (were?) famous for not playing very nice when they had
to share interrupts with other hardware. Best practice is (was?) to have
the card on its own interrupt. Did you try to disable usb on the box and
disabling the uhci_hcd & ehci_hcd modules? What about disabling the
parallel port and the parport0 module (to free up as much interrupts as
possible)?

Since I see ide0 in your output check if you have set the right hdparm
settings for IDE. There are some tips on
http://www.asteriskguru.com/tutorials/pci_irq_apic_tdm_ticks_te410p_te405p_noise.html

Is this box using Intel nics and the e100 module? There seem to be some
latency issues with the e100 module. Try the eepre100 module instead of
the e100 to see if it makes a difference. See this posting to lkml:
http://www.gatago.com/linux/kernel/6130880.html

Regards,
Patrick





More information about the asterisk-ss7 mailing list