[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference azur::mcc

Title:DECmcc user notes file. Does not replace IPMT.
Notice:Use IPMT for problems. Newsletter location in note 6187
Moderator:TAEC::BEROUD
Created:Mon Aug 21 1989
Last Modified:Wed Jun 04 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:6497
Total number of notes:27359

3904.0. "Ultrix mcc install, dce/threads, core dump" by ZPOVC::RAMARAJ () Thu Oct 15 1992 11:25

Had this problem when installing DECmcc �BMS on Ultrix.

Ultrix V4.2a
24MB memory
665Gb disk
DS5000 200
DEC threads in the BMS kit has been installed w/o any error
All config for kernel(param.c, sem.h...) and patches for vm_proc.o have 
been done.

At the enrolling point

.....
enrolling mcc_alarms_fm

exception. Invalid memory address (dce/th)
usr/etc/subsets/MCRBMS120.scp: 1033 abort - core dumped

I tried enroll manually again but got 
segmentation fault

Any ideas on what is causing this.

Raj
DEC Singapore
T.RTitleUserPersonal
Name		DateLines
3904.1How about swap space?TOOK::MINTZLKG2-2 near pole X3, cube 6072, dtn 226-5033Thu Oct 15 1992 11:284
How about swap space?  Check with "pstat -s"

Is this the first installation, or did you have a field test version
on the same system?
3904.2100MB and deinstalled FT kitZPOVC::RAMARAJThu Oct 15 1992 11:456
    Swap space is 100MB.
    
    I had a FT version earlier.  Deinstalled it completely and removed all
    the directories /usr/mcc and /var/mcc.
    
    Raj
3904.3Time for a QARTOOK::MINTZLKG2-2 near pole X3, cube 6072, dtn 226-5033Thu Oct 15 1992 11:577
It sounds like you have done all the right things.

Is it only the alarms FM that fails to enroll?
Can you enroll other MMs manually?

Since it doesn't sound like this is a known problem, you should
probably file a QAR (see note 7).
3904.4All fm's same problemZPOVC::RAMARAJThu Oct 15 1992 12:296
    All fm's have the same problem,  I cannot enroll any of it.
    
    All give segmentation fault.
    
    Could you QAR, please.  No access to qar system, never quared before.
    Raj
3904.5See note 7 for how to QARTOOK::MINTZLKG2-2 near pole X3, cube 6072, dtn 226-5033Thu Oct 15 1992 13:4721
The QAR system is now generally available, giving everyone access
to track their own QARs.  See note 7 for instructions.

Meanwhile, please do the following so we can start investigating:

		setenv MCC_LOG  0x40000

Then re-enroll some module and mail me a pointer to the core file.

Also, can you issue the command:

		% manage
		MCC> show mcc 0 all char

and include the results in the QAR.


Thanks,

-- Erik (TOOK::MINTZ)
3904.6Maybe MIR table are corruptedTAEC::LAVILLATThu Oct 15 1992 14:0924
Re .4:

>
>    All fm's have the same problem,  I cannot enroll any of it.
>    
>    All give segmentation fault.
>    
>    Could you QAR, please.  No access to qar system, never quared before.
>    Raj
>

	I think this is an known (QARed) problem of MIR routines.

	try to remove your mir repository table via :
	
		rm /var/mcc/mcc_mir_repo*

	and try to enroll again.

	Hope this helps.

	Pierre.
3904.7TOOK::SWISTJim Swist LKG2-2/T2 DTN 226-7102Fri Oct 16 1992 09:274
    What known problem is it?  I don't know of anything that can cause
    this except maybe an old field test version of a repository table
    still in /var/mcc.
3904.8Red herring alertTOOK::MINTZLKG2-2 near pole X3, cube 6072, dtn 226-5033Fri Oct 16 1992 10:287
From .2:

>    I had a FT version earlier.  Deinstalled it completely and removed all
>    the directories /usr/mcc and /var/mcc.

It does not sound like old repositories are the problem.
3904.9Will get log file on monday.ZPOVC::RAMARAJFri Oct 16 1992 11:515
    Eric,
    sorry haven't had the time to get the log output.  Its at another site.
    Will get on Monday.
    
    Raj
3904.10Was talking of MIR_INIT_FAIL pb.TAEC::LAVILLATFri Oct 16 1992 13:5520
Re .7:
>
>    What known problem is it?  I don't know of anything that can cause
>    this except maybe an old field test version of a repository table
>    still in /var/mcc.
>    

	Jim,

	It is the MIR_INIT_FAIL problem that can cause this : once you
	have MIR_INIT_FAIL each MM enrollment will fail with segmentation fault
	you have examples of corrupted MIR repository tables under :

	directory UNITEL::/kits/mcc 
	files mcc_mir_repository_table.pag and mcc_mir_repository_table.dir

	Regards.

	Pierre.
3904.11TOOK::SWISTJim Swist LKG2-2/T2 DTN 226-7102Tue Oct 20 1992 08:466
    OK, Pierre.  I didn't make the connection between the INIT_FAIL problem
    (which we still can't find) and this problem.
    
    If what you are saying is true, why didn't we see a MIR_INIT_FAIL
    earlier in this guy's installation?
3904.12PM is not notified of the errorTAEC::LAVILLATTue Oct 20 1992 10:0315
>    
>    If what you are saying is true, why didn't we see a MIR_INIT_FAIL
>    earlier in this guy's installation?
>    

	Because it is the MM process that exits with MIR_INIT_FAIL, and
	the PM is not notified of this status. The response does not
	arrives, then the PM tries to break the RPC connection, and die
	with segmentation fault.

	BTW was it really the problem ?

	Regards.

	Pierre.
3904.13TOOK::SWISTJim Swist LKG2-2/T2 DTN 226-7102Tue Oct 20 1992 16:515
    Pierre,
    
    OK, we've fixed the bug that causes a seg fault if the target MM
    fails to enroll (1.3), and I'm copying one of your corrupt repo files
    right now to look at it.
3904.14V4.3 installed fresh, ok nowZPOVC::RAMARAJThu Oct 22 1992 11:1710
    I upgraded the ultrix to V4.3, installed fresh and installed MCC.
    All ok now.  
    
    I'm not sure what caused the earlier problem, but because we had to get
    it up quickly, I tried on my spare disk first with v4.3 ultrix and did
    it on the customer's disk.
    
    Hope the problem with the segmentation fault has been rectified.
    
    Raj