Conference csc32::consolemanager

1530.0. "Corrupt Console Manager files" by KERNEL::BARKERH () Wed May 07 1997 10:05

Hi,

I have a customer running PCM 1.6 ECO 3 (VMS 6.1) who sometimes gets his 
console manager file corrupt and thus gets the following:

$ CONSOLE VERIFY CONSOLEMANAGER

        Starting Host ConsoleManager

Loading : CONSOLE$LOGFILES:ConsoleManager. fileset
          LOG file
Test 1, Check to ensure that there are the same number of timestamp
        records in the .TIMES file as there are <NL> characters in
        the .LOG file, there should be a one-one match.
Test 1 : ** FAILED **, Too many lines of data for available timestamps
              <1178088> Timestamps in the TIMES file
              <1292126> Records found in the log file
Test 2, Check to ensure that the first timestamp record points to
        offset 0 in the .LOG file
Test 2 : Passed.
Test 3, Check to ensure that all our timestamp records point
        somewhere within the logfile and that they point to
        newline characters
Test 3 : ** FAILED **, Time record 18990 does not point at <LF>
              Time record 18991 does not point at <LF>
              Time record 18992 does not point at <LF>
              Time record 18993 does not point at <LF>
                .  
		.
                .

Note usting /REPAIR makes no difference!


Also, a $CONSOLE ARCHIVE CONSOLEMANAGER/KEEP=72 just goes in to a loop and
stops another user being able to archive.


Anyone explain why this happens and whether there's a workaround?

Thanks,

	Henry (Basingstoke CSC)

    Hmmm. I can explain the loop as I've seen it and analyzed it. What
    happens is the .EVENTS file manages to get some junk between records
    and cannot get past it. What causes this is typically something
    extraneous such as a user making a copy such that the end-of-file
    pointer get updated to be the last byte in the file. When PCM opens the
    file for append access and starts writing records, you now have junk
    between the records. Now, in older ECO levels of PCM there were some 
    bugs uncovered in logging. It was required that you rename or 
    archive the old logfiles and allow PCM to create new ones after
    installing at least ECO 3.
    
    Since you already there, I recommend that you shutdown PCM, rename the
    old CONSOLEMANAGER.* files and restart PCM. The parent daemon will
    create noew ones and at this point you need to sit back and monitor
    things. The customer can reattmept the archive of consolemanager 
    after a few days. Makre sure they aren't using somehting other than a
    PCM interface to manipulate the logfile sets.
    
    REgs,
      Dan