dump space error

[Ravi_8]

Hi,
In a D390 HP box we had 11.0, system started hanging, after so many tries we couldn't bring it back and decided to reinstall the 11.0.
while installing it started giving the following error:

****A system crash occured
**** system is now preparing to dump physical memory in to disk for debugging the crash

error:No dump space is configured, physical memeory will not be dumped.

(we are trying to install from CD)

help pls...

thanx in advance

[Vincenzo Restuccia]

Verify SCSI termination and PIN of cables.

[Bill McNAMARA_1]

Disconnect all connected hardware periferals and try the install again.

Create an ignite image/ tape from another D class and install from ignite tape.

Dump is usually lvol2 in vg00 or you can dump on disk by modifying the kernel.. but seeing as you're in an install, it's a little more difficult.

You may have a h/w failure.. so the best avenue would be to bring the system down to a minimal state and add h/w gradually until you notice the pb..

Just before the dump is taken, what messages are on the screen?

Good luck
Bill

[Ravi_8]

recently one HP engineer came and upgraded the RAM(256 MB to 512) and procesor(single to dual) and swap from 512 MB to 1024 MB. it worked well for nearly months and then started hanging, and while installing now it is giving the above mentioned error. is the upgradation causing this problem? where as 10.2 in another HDD in the same system is up and running.

[Bill McNAMARA_1]

The dump is triggered because of a panic or HPMC. You need to get the message before the dump.

Stop at BCH (you got 10 seconds on boot up to interrupt the boot process..
the type ser
and
pim hpmc

send us the output to determine what h/w has failed.

You know the weimar distribution curve? If the h/w doesn't fail in the first 6months.. it probably isn't going to......

Later,
Bill

[Ravi_8]

ser> pim hpmc gives the following

processor 0 HPMC Information
Time stamp = Mon Apr 16 07:27:21
HPMC chasis code = 0xcbf0 0x5008 0x5408 0xcbfb
Generic register 0-31
00-03 0000000000000000 0000x3fg0000 ....
04-07 0000x00300005000 0007x0560900...
08-11....
.
.
28-31 00......

control register 0-31
00-03 0000000000000000 .......
.
.
28-31 000000030x000000 ....

space register 0-7
00-03 00000000 0dd33c00 00000000 00000000
04-07 00000000 00409c00 00509400 00000000


and then comes back to prompt

[Bill McNAMARA_1]

FLT CBFB Branching to the OS HPMC handler
FLT CBF0 HPMC handling initiated
FLT 5xy8 Processor Memory bus broad fault

You've got how many cpu's? CPU 0 and 4 look okay from the pim..

some doc I found to describe this: FLT 5
5X08. BROAD_ERROR received by this processor:
NORMALLY: This CPU did not detect the problem, and may or may not have CAUSED the original problem. Focus on logs of other CPUs, IOAs, or Memory..

Broad_error indicates that SOME MODULE on bus 0 (Runway) has detected (and reported) a serious error on the bus. All modules (except memory) which do not have previous hard or fatal errors logged will log a hex code indicating "Broad_Error".. The module that signaled the Broad_Error is a different module (i.e. not THIS CPU), so the cause is normally found either in the logs for another processor or in the logs for the IOAs or memory.

[Bill McNAMARA_1]

can you send the output of SL from BCH, to get a history of the hex code o/p.
The full pim is decodeable to identify specific component... specifically the last line identifier and responder. You may want to send that on you your rep.

Bill

[Ravi_8]

Hi,

bill thanx a lot for u r reply.
i changed the target HDD (i,e from 8/4.10 to 8/4.9) and installation was smooth. and extended the vg to 8/4.10, i didn't get any problem.

later
ravi

[Ravi_8]

hi

it's started hanging again..

any solutions...

[Ravi_8]

hi,

i couldn't find 'tombstones' files anywhere in the system. i am attaching the portion of the log file which i looks suspect.

thanx

[Bill McNAMARA_1]

some good news at least,
I find it suspect that the disk you originally selected does not install.. is the capacity the same as the one that it can install too?

when does the system hang, at vg00 activation on system boot or randomly while you're logged in?
The log file you attached shows nothing suspicious.. All your memory is identified and swap and dump are defined.

Your syslog may give some more information if the memory you had recently installed is giving problems, such as single bit errors..
but in cases like that with real mem pbs you'd end up getting panics.. not hangs.. hangs are more io related.

You might want to get a repetitive sar output and redirect into file during the system operation until the hang, reboot and find out from the sar o/p if there are io or cpu problems at the hang time.

You might want to get a cfg2html O/p of your hardware whilch will also gather log files etc.. to a html file. it might make the analysis easier. come.to/cfg2html

Relating to the FLT code, a broad error is caused by some element, ie a CPU detecting an uncorrectable parity error and letting all elements on the bus, know that an error occured with a broad_error signal. Identifying from this code you can find out which element detected the parity problem and where it came from. This unfortunately is complicated stuff, and you need documentation about the chip registers to identify.. as well as knowledge of the D class in detail. slot 0 on the system bus is CPU, and slot 4 I believe is memory.. the rest of the hpmc is the registers that allow detection of individual components of the cpu registers at the time of the panic to debug source cause. Really only HP can do this...

Fortunately or Unfortunately too, I'm on holls for an entire month and will try to actually profit from them.. sorry I'd have liked to have helped more with the solution...

Later,
Bill

[Ravi_8]

the disks are of the same size, and it hangs randomly, let me contact HP .

thanx bill

have a nice holidays

later
ravi