Conference help::dns

1439.0. "Resource problem ?" by PRSSOS::CAPAZZA (Jean-Denis CAPAZZA, NSAP-TSAP-SSAP-PSAP...ASAP) Mon Apr 14 1997 06:05

	Hi,
	
	Does somebody have an idea about the folowing BUGCHECK.
	It seems to be a resource problem, but which one ??

	DNS V2.0 ECO 70

	Thanks for any help
	Jean-Denis

$ typ DNS$SERVER.LOG;140
-----------------------------------------------------------------
-- Log for mvax17$dka700:[sys0.syscommon.][sysexe]dns$server.exe;4 (pid 
2739) op
ened Tue Feb 11 10:14:44 1997
-- Digital DECdns Server V2.1-SSB-070
--     ECOs applied: 1 through 70
--     SERVER_MAIN built at:  Dec  2 1996 13:43:59
-----------------------------------------------------------------
Tue Feb 11 10:15:36.485:   Transaction Log playback statistics: 189c3c->19-
OCT-1
993 06:32:50.42/AA-00-04-00-1D-FC
Tue Feb 11 10:15:36.485:     Started:  Tue Feb 11 10:15:36 1997
Tue Feb 11 10:15:36.605:     Finished: Tue Feb 11 10:15:36 1997
Tue Feb 11 10:15:36.605:     Creates: 0 Directories
Tue Feb 11 10:15:36.605:              0 Objects
Tue Feb 11 10:15:36.605:              0 Do_Updates
Tue Feb 11 10:15:36.605:     Updates: 872 to directories
Tue Feb 11 10:15:36.605:              0 to clearinghouses
Tue Feb 11 10:15:36.605:              670 to objects
Tue Feb 11 10:15:51.435: ************* BUGCHECK *************
A module in the server has bugchecked
Bugcheck code is ... -902
Thread id .......... 144638
Thread errno ....... 0
errno .............. 2
Module Name......... BLD$LOCAL:[SRC]DB_COMPRESSION.C;9
Module Line #....... 1055
Module build time... Dec  2 1996 13:11:56
Last status code ... -999
Tue Feb 11 10:15:51.455: ************* BUGCHECK *************
At 8522, writing up to last 500 allocations to vm_data
At 8522, writing all current allocations to vm_data
GET_VM calls: 5574  FREE_VM calls: 3020  VM  allocated: 10050480 bytes
Process FREPTECNT: 124408 - Number of pages avail for virt mem expansion
Process PAGFILCNT: 127053 - Remaining paging file quota of process
Process PAGEFLTS: 23934  - Total number of page faults incurred by process
Process WSSIZE: 29502 - Current working set size of process
Process BIOCNT: 198 - Remaining buffered I/O quota
Process BYTCNT: 49824 - Remaining buffered I/O byte count quota
Process DIOCNT: 18 - Remaining direct I/O quota
Process FILCNT: 94 - Remaining open file quota (includes DECnet links)
Process FILLM: 100 - Open file limit quota
Process PAGFILLOC: 68157440 - Current paging file assignment (4th byte is 
pg file index)
Process PGFLQUOTA: 150000 - Paging file quota of process
Process WSEXTENT: 40004 - Current working set extent
Process WSQUOTA: 13337 - Working set size quota
System CHANNELCNT: 127 - Permanent I/O channel count
System FREE_GBLPAGES: 32224 - Current number of free global pages
System FREE_GBLSECTS: 386 - Current number of free global section table entries
System NPAGEDYN: 4999680 - Non-paged dynamic pool (bytes)
System PAGEDYN: 2341376 - Paged dynamic pool (bytes)
System PAGEFILE_FREE: 170406 - Number of free pages in currently installed 
paging files
System PAGEFILE_PAGE: 173608 - Number of pages in currently installed 
paging files
System SWAPFILE_FREE: 19096 - Number of free pages in currently installed 
swap files
System SWAPFILE_PAGE: 19096 - Number of pages in currently installed swap files
System VIRTUALPAGECNT: 150000 - SYSGEN param: no. of pgs that can be mapped 
for a proc.
System WSMAX: 40000 - system wide max size of a process working set 
regardless of process quota.
$

    This is not related to a resource problem. This is an assertion failure
    in the db_compression routine that writes the checkpoint file. In this
    case the nodes in do not equal the nodes written.

    This problem rarely occurs and I would suggest that you ignore it
    unless it become more frequent.

    Regards,
    Bob Wheater
    DECdns Engineering

	Hello,

	we need a workaround, if possible, because at this customer site, there 
   are five DNSserver. And on two of these servers this BUGcheck occur at each 
   DNSserver STARTING. 
	These two DNSserver are down.

	THANKS IN ADVANCE.
	REGARDS
	JEAN-LOUIS MONZON.

Hi Jean-Louis:

    Please run the surgeon over the checkpoint files of the servers that
    are bugchecking. Use the -scanrx and -scand options.

    Check for the word "corruption" in the -scanrx output. Also check
    that the number of directories in the -scand output is equal to
    the number of directories in the header.

    You may want to use the lastest version of the surgeon:


	dragns::user$91:[wheater.public]dns_db_surgeon.c


    compile is using VAXC (as we have had problems with decc in the past).


    If you have corruption you can remove the bad directory with the
    "excised" command or is the directory count is bad it can be adjusted
    by using the "changed" command.

    if you have neither  problem or don't understand how to correct the above
    problems you should submit an IPMT.

    Regard,
    Bob Wheater

                Hello Bob,

        I ran SURGEON process over the checkpoint files of the servers that
  are bugchecking using the -scanrx and -scand options without problem.

        I will submit an IPMT.

        Regard,
        and many thanks for your help.

                                JEAN-LOUIS