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

Conference aosg::lsm

Title:LSM
Moderator:SMURF::SHIDERLY
Created:Mon Jan 17 1994
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:803
Total number of notes:2852

788.0. "Why LSM/HSZ40 resync at boot time" by 22740::TERENCELEUNG () Wed Apr 30 1997 10:45

    Our customer has two 8200 with 2 pairs of HSZ40 located in different SCSI
    bus.  There are 18 RZ29B-VA sitting on each pair of HSZ40.  We
    implemented HSZ40 RAID 5.
    
    LSM is used to mirror this two pairs HSZ40's data.
    
    The problem is Whenever we reboot the system, these two pairs data
    takes several hours to resync.  And we got the following messages:
    
    ...LSM: Resynchronization of volume XXX in ....
    		.
    		.
    		.
    Our customer just concern whether it normal or not.  Can any LSM
    guru give us an explanation in deep detail.
T.RTitleUserPersonal
Name
DateLines
788.1clean stop ?BRSDVP::DEVOSManu Devos NSIS Brussels 856-7539Thu May 01 1997 11:1311
    If LSM is re-synchronizing volumes, this is because these volumes have
    not been cleanly stopped at the shutdown time. So, something is not
    correct in the way you shutdown the system. Are you using init 0 which
    stops everything cleanly or some more abrupt command (like halt,
    reboot, shutdown -r ).
    
    So, tell us how you reboot the system and if you stop the application
    using the HSZ disks 
    
    Manu
    
788.2We tried both shtdown -h & shutdown -rPRCFS1::SOCHUNWAHFri May 02 1997 02:0713
    
    Hello,
    
    	We had tried both "shutdown -h" and "shutdown -r" and have the same
    results.  
    
    	We'll try "init 0" next time to see any different, but because next
    two days is holiday.  Probably the test will be performed next Monday,
    and will update the results here.
    
    
    SO
    
788.3BRSDVP::DEVOSManu Devos NSIS Brussels 856-7539Fri May 02 1997 19:399
    So,
    
    You should use "init 0" and not hte old Ultrix "shutdown -r".
    
    Init 0 is running the scripts in /sbin/rc0.d which is  stopping
    everything.
    
    Manu
    
788.4Now I'm confused...DYOSW5::WILDERDoes virtual reality get swapped?Mon May 05 1997 11:3210
    Well, I'm confused....I have seen notes stating that you shout use init
    S, init 0, AND shutdown -r. Maybe engineering can give us a definite
    answer. 
    
    What is the CORRECT way to shut down a cluster system and re-boot it?
    
    Thanks,
    
    /jim
    
788.5BRSDVP::DEVOSManu Devos NSIS Brussels 856-7539Mon May 05 1997 17:1219
    Jim,
    
    I understand your confusion...
    
    At the beginning (V1.0) ASE requested "shutdown -h".
    Now, ASE is really happy with "init 0" which properly run the service
    stop script(s) and stop the ASE daemons and AM Kernel driver.
    
    LSM also is needing "init 0" to allow the "clean stop" of all volumes.
    A volume which has not been stopped before the reboot will be
    resynchronised.
    
    And about the "init s, init 0 ", maybe you are referring to a note for
    the Trucluster software, which explained that we cannot use the
    command "init s" followed by "init 3" because the member will NOT join
    the cluster properly.
    
    Manu. (Not from Engineering:-)