D410P and HDLC errors

Hello,

I've installed a D410P card and configured its ports as E1 PRI. It works fine
for a few minutes (10-15) handling calls and then suddenly it reports HDLC Abort and/or HDLC Overrun and all spans go down. The link status is active but HDLC/PRI layer is not. On HDLC level it's sending SABME requests but getting no response. When I restart Asterisk, it provisions all ports without any problem and HDLC link is up. I disabled all unneccessary devices and it seems no interrupts are shared and irq misses are repored.
I am running Asterisk 1.4.19 and Zaptel 1.4.9.2 - maybe I should switch to 1.2.x?

Has someone any ideas how to solve the issue?

Best regards,
Michal

hello, Michal:
as i know, the zaptel-1.4.9.2 is buggy version. maybe you can downgrade from the current version to asterisk-1.4.18 and zaptel-1.4.8 and try it.
Regards!
James.zhu

What is your cpu/motherboard/ model? and which linux you are using?

Please switch S7.2 to off to have a test.

I'll try to downgrade to 1.4.18 (and if it still fails) to 1.2.x as well as the S7.2 DIP set to OFF and let you know about results.
The failing scenario goes as follows:
1. Eveything works well for a few minutes.
2. HDLC Overrun and/or HDLC Abort are signalled.
3. PRI spans go down (HDLC/signalling layer).
4. HDLC SABME is sent, but neither response nor SABME from remote ends
are received. Complete restart of Zaptel/Asterisk usually helps.

Switching PCI Burst DIP to OFF solved the problem with HDLC errors.
Although next problem popped up - when new trunks are plugged in,
they go up and down, up and down and refuse to be stable. After one
or two reboots they start to work fine and when they successfully reach
a "stable" state, everything works fine.

Just to let know - after peekpoking a while I downgraded my asterisk/zaptel to versions 1.4.17/1.4.7.1 (iit seems to be a good/stable combination) and everything started to work fine. Also changed CPU from E2200 to E4700.