topic Re: rx2600 with cache problem, won't boot in Integrity Servers

rx2600 with cache problem, won't boot

Unixnerd — Tue, 11 Mar 2008 12:32:11 GMT

Let me start by saying that I have about 20 years experience with PA-RISC workstations but this is the first Itanium server I've touched.

I'm getting the following errors:

Alert Level 3: Warning
Keyword: MEM_CACHE_LINE_0_WR_RD_FAILED
Memory ECC normal write/read test failed

Alert Level 3: Warning
Keyword: BOOT_CPU_LATE_TEST_FAIL
CPU failed late selftest

Alert Level 3: Warning
Keyword: BOOT_CPU_CONFIG_FAIL
CPU struct init failed
Logged by: System Firmware 1

When I then try to run a CO command from the monitor to start a console on the main processors it fails (no shock there!) and hangs until I press Ctrl-B to return to the monitor cpu.

I've tried physically swapping the two 1.06GHz cpus over but it makes no difference, is this referring to an external cache I wonder?

The machine beeps twice on power up, but I don't think that's related.

Could this actually be a problem with the main (12Gb) ram? I've reseated all of it.

Thanks in advance :-)

Re: rx2600 with cache problem, won't boot

Sameer_Nirmal — Tue, 11 Mar 2008 14:57:21 GMT

The problem appears to be with one of the DIMMs of the main memory.

Here is the OS event detail which is the same as reported by SFW/MP in their own format.

Event 299

* Severity: MAJOR_WARNING
* Event Summary: Memory ECC normal write/read test failed
* Event Class: System
* Problem Description:
After FW's first access to main memory, FW detected that the CEC logged an error after reading back what was just written.
* Cause / Action:
C: The DIMM that maps to cache line 0 is in a chipspare condition A: Contact HP support C: The DIMM that maps to address 0 is not seated properly A: Check all of the DIMMs in the system and make sure that they are inserted fully into the slot with the retention mechanism in place C: System may be running at the wrong frequency. A: Verify the system bus frequency and the memory bus frequency.
* Automated Recovery: None
* Event Generation Threshold: 1 occurrence

The DIMM problem caused the CPU test failure which in turn made it "deconfigured".

Re: rx2600 with cache problem, won't boot

Sandy Chen — Tue, 11 Mar 2008 15:27:56 GMT

Hi,

Does the 12GB of memory just been added? Perhaps you could try with minimum config for the memory, just a quad to make sure your memory is the culprit.

Regards,
Sandy

Re: rx2600 with cache problem, won't boot

Unixnerd — Tue, 11 Mar 2008 17:50:01 GMT

Thanks guys, it was a bad simm! When the error message said CACHE I just assumed it was cpu cache.

Can't thank you enough.