EMS Failure

[Mike Linger_2]

Ok. I've still got an EMS failure on all three of my rx5670s.

I see p_client. What other processes should be running? As I recall, our HW configuration should support cmc_em, disk_em, ia64_corehw, memory_ia64, and sysstat_em.

I can't even verify an existing monitor:
/opt/resmon/bin/send_test_event -v ia64_corehw

This verify test used to work.

I see a lot of messages in the log files that lead me to believe port communication failure.

I recall one of our engineers modifying system latency timers to improve the performance of his application.

Is there a way to set port latency to zero (system wide)? If so, how do you do this? What is this value supposed to be? How do I restore this value to its default?

[Eugeny Brychkov]

Usually running diagmond daemon is a sign of EMS. Check file /etc/rc.config.d/diagnostic to contain DIAGNOSTICS=1 and then start diagnostics with
/sbin/init.d/diagnostic start
Eugeny

[Mike Linger_2]

Looks like I can start or stop at will, but this doesn't fix the problem.

Could this be the result of our firmware/processor upgrade?

# /opt/resmon/bin/send_test_event -v ia64_corehw

Finding resource name associated with monitor ia64_corehw.

Found resource name /system/events/ia64_corehw
associated with monitor ia64_corehw.

Creating test file /var/stm/config/tools/monitor/ia64_corehw.test
for monitor ia64_corehw.

Performing resls on resource name /system/events/ia64_corehw
for monitor ia64_corehw to cause generation of test event.
Contacting Registrar on cleuit3
resls: Cannot contact Registrar on cleuit3
/opt/resmon/bin/send_test_event: Failed to perform resls on resource name /syste
m/events/ia64_corehw for ia64_corehw to cause generation of test event.
System call failed with return value 256.