1748251 Members
3673 Online
108760 Solutions
New Discussion юеВ

987/200 LPMC's

 
SOLVED
Go to solution
Fred Metcalf
Trusted Contributor

987/200 LPMC's

We are getting these LPMC's on a 987/200. Our local CE's cannot find the source and state that LPMC's are self-correcting errors and can be ignored. On Dec 28 2000 we had a System Abort, no console message. System was restarted and is running OK. These were the CPU Display error codes: FLT DEAD 01E2 0200 BF00 No memory dump possible, TC resulted in FLT 2040 CBFF.
Can I ask for a second opinion?
More Extracts from the system log files in the attachment.
===============================================================================
THU, JAN 4, 2001 7:35 PM LOG2661.PUB.SYS SYSTEM (PIN 226)
LPMC
HMODEL: $287 HPA: $FFFBE000
PIM LENGTH: 44 CHECK TYPE WORD: $0
VIPER STATUS REG: $340 CACHE CHECK WORD: $0
TLB CHECK WORD: $0 BUS CHECK WORD: $0
ASTS CHECK WORD: $0 ASSIST STATE WORD: $0
SYS. RESPONDER ADDR:$11993E30 SYS. REQUESTOR ADDR:$0
===============================================================================

Missing MPE :-)
3 REPLIES 3
paul courry
Honored Contributor

Re: 987/200 LPMC's

Well here goes, a LPMC is a Low Priority Machine Check which IS a self correcting error, but it Is Not A Good Thing.


FYI, a TC can die and make it impossible to do a memory dump. I've had it happen to me, but don't stop trying!

Given the number of errors I would say the following words to the CE "Sir, I am formally requesting an escalation of this problem." and back it up with a phone call to the RC noting the open hardware call number you have. This will get additional resources thrown at the problem and while the CE will still be the lead he is NOT in charge of the escalation.
Eric Sorensen_1
Honored Contributor

Re: 987/200 LPMC's

Hello Fred,

It is true that LPMC's can be self-corrected and ignored in most cases, but a high number of LPMC's suggest something more serious. I suggest looking at the recent logfiles to learn just how many LPMC's are logged (sysdiag logtool summarize and look for type #152). Also look at LOGTOOL>memrpt. Sometimes the cause of the LPMC's can be cleared with a shutdown and power-cycle, but if not, then I would request the hardware folks do memory diagnostics.
A problem well defined is half solved.
Kwan Fong, Emile
Honored Contributor
Solution

Re: 987/200 LPMC's

The key thing here is the second fault code:
FLT 2040 CBFF.

2040 indicates a Data Cache Parity Error
CBFF indicates an HPMC.

Cache error occur on processors boards; since you have 2 processors (s987-200), I would recommend you get an HPCE onsite and check the PIM area to find out which processor is generating the data cache parity errors and replace it.

To check the PIM area for s9x7 servers:
1) Boot with TC from the CM> prompt.
2) Boot from primary? N
3) Boot from alternate? N
4) Enter boot path, command, or? PIM
King is the customer!