Conference azur::mcc

5366.0. "DECmcc V1.3 MIR corruption" by TAEC::WEBER () Wed Jul 21 1993 03:22

    Hello,

    Please find  hereafter the descrition of a MIR file problem as reported
    by a customer.
    This problem has also been identified by other french customers.
    Do you have advises to prevent this problem to happen ?
    Thanks for an answer,

    Florence

    
-- Submitted by --                      -- DIGITAL contact --
EUAN BARKER, DIGITAL, BRISTOL, UK       EUAN BARKER                             
                                        BRISTOL (BSO)                           
                                        (44)-454-623427                         
                                        
                                        

Attachments: NONE                      		Customer severity: H
Reproducible at will: N

CPU        Memory     System device
CPU RISC   16Mb                 

Hi,

	This is an occasional problem but is serious in nature. It may not
apply to TeMIP only but be a generic mcc problem, but needs following up either
way :-

	The MIR's seem to become corrupt occasionally on the xxxxxxxx system.
This has been happening for a while and we have tended to put it down to
the fact that they are developing on this system so things may become corrupt
with people hacking away etc. What xxxxxxxx have done recently is all use
their own MIR files to try and reduce the risk and impact when this happens.
It is still happening, below are details of the latest :-

The system was running a demo of the various am's xxxxxxxx have developed
with mcc and TeMIP :-

demo processes :

ps -x

mcc_notification_fm 10 N
mcc_domain_fm 6 N
temip_alarm_handling_fm 18 N
mcc_bsc_am 22 N
mcc_bts_am 26 N
mcc_msc_am 29 N
mcc_hlr_am 28 N
mcc_css_am 34 N
mcc_omcr_am 31 N
/usr/users/demo/mmexe/n_sk_msc 89 6
/usr/users/nms/CURRENT/cm/n_cm0500_x25 4250 X25-C X29login
mcc_registration_fm 12 Y

DECmcc error:

mcc_fcl
MCC> show msc msc.testbed all ref

MSC LOCAL_NS:.msc.testbed
At 1993-07-16-11:43:53 References

The requested operation cannot be completed
	MCC Routine error = %MCC-F-MIR_INIT_FAIL, the MIR could
				not be initialised

DECmcc environment variables :- (opnly ones different to standard ):-

MCC_MIR_LOCATION=/usr/users/demo/mcc_system
MCC_MMEXE_LOCATION=/usr/users/demo/mmexe
MCC_SYS_LOCATION=/usr/users/deno/mcc_system
MCC_MAPS=/usr/users/demo/maps
MCC_DICT_CACHE=

ls ~demo/mcc_system

dap.cmd
mcc_dap_keytbl.txt
mcc_dispatch_table.dat
mcc_fcl_keytbl.txt
mcc_fdictionary.bpt
mcc_fdictionary.ctm
mcc_fdictionary.dat
mcc_fdictionary.dat.dir
mcc_fdictionary.log
mcc_impm_expand_field.mcc_temip_nms_pm
mcc_meta_definition.dat
mcc_meta_dictionary.dat
mcc_mir_repo_table.dir
mcc_mir_repo_table.pag
mcc_target_data_mir.dir
mcc_target_data_mir.pag
mcc_target_inst_mir.dir
mcc_target_inst_mir.pag


Problem currently resolved by performing an mcc_kill, removing the
MIR Repositry table (mcc_mir_repo_table.dir/pag) and re-enrolling
the required MM's.

Hope you can help,

	regards,

		Euan

    The next time this happens, we need to capture the corrupt master
    repository.
    
    Could you put a copy of the corrupt mcc_mir_repo_table.pag somewhere where
    I can copy it via ftp (or else uuencode it and mail it to me).
    
    (I don't need the .dir file).
    
    The easiest way to do this is to simply rename both the .dir and .pag
    files to something else as soon as you see the error.  Then the system
    will create new ones at the next restart and you can continue while
    conserving the corrupt file.
    
    Note that the last time this problem was reported, it was also from
    TEMIP.  Pierre Lavillat mailed me the file and I found that the
    corruption was due to a text message from the malloctrace tool having
    written over a block of the repo_table .pag file.  You might check
    to see that this is not still the case (dump the file in ascii using
    od(1) and it will be obvious).
    
    Jim
    

    Jim,
    
    	I've got a copy of the corrupt mir on the easynet now, Sorry for
    the delay. It's available on ::--
    
    	briris::'/more-users/euan/*mir*'
    briris is DEcnet address 44.209, ftp may work, 16.182.144.100
    
    The two files there are the ones you requested, with .cor extension to
    avoid any confusion.
    
    Regards,
    
    		Euan