Operating System - OpenVMS
cancel
Showing results for 
Search instead for 
Did you mean: 

quorum connection lost and then reagained - normal?

 
SOLVED
Go to solution
Homer Shoemaker
Frequent Advisor

quorum connection lost and then reagained - normal?

 

System is an rx2660 OpenVMS 8.4 using a shared system disk on P2000 MSA via fiber channel.

 

It is part of a two node cluster with a quorum disk being a separate volume on the shared storage.

 

(The other node is a disconnected member.  For now,  it is booted on a local drive as a stand alone production server.  Current production data is also still on one of it's local drives.)

 

 

The attached text file is captured from the console display  - from shutdown until the reboot is complete.

 

I have two questions about messages that come up during bootup. 

 

1) At about line 150 in the attached text file you will see that Ithe quorum is lost and then regained.  I'm not thinking that's normal.

 

2) When I mount the quorum disk I get a dismounted improperly message and notified that a rebuild is in progress.

 Here is how and when I mount the quorum disk. This is at the bottom of systartup_vms.com.

 

 

$!******************************************************************************
$!     **** MOUNT PUBLIC DISKS ****
$!******************************************************************************
$!     ( $1$DGA102 IS THE CLUSTER QUORUM DISK )
$!******************************************************************************
$ MOUNT/SYSTEM/NOASSIST $1$DGA102: DMS_PROGRAMS DMS$PROGRAMS
$ MOUNT/SYSTEM/NOASSIST $1$DGA103: COMPANY_DATA CMP$DATA
$ MOUNT/SYSTEM/NOASSIST $1$DGA104: USER_DATA    USR$DATA
$ MOUNT/SYSTEM/NOASSIST $1$DGA105: DEVELOPMENT  DEVELOPMENT
$!

One other thing... I don't know if this is pertinent.  Besides being the quorom disk, the other thing that is different about the quorom disk is that I did wind up initializing it and renaming the volume more than once while configuring the cluster software.

 

It took me a few trys to get the sequence of how things ought to be done correct.  (At least I think I finally got it correct.  Either or both nodes can join the cluster.)

 

Any help with these question or any other comments would be appreciated.

 

Thanks

3 REPLIES 3
Volker Halle
Honored Contributor
Solution

Re: quorum connection lost and then reagained - normal?

Homer,

 

after a cluster state transition, the system  temporarily releases and re-acquires the quorum disk - this is normal behaviour.

 

As the quorum disk is not being dismounted cleanly during shutdown, consider to mount it using /NOREBUILD.

 

If you build a new cluster with a quorum disk, consider setting EXPECTED_VOTES=1 first and boot the first system as a single-node cluster while DISK_QUORUM is defined. Then mount the Quorum disk /SYSTEM and the quorum.dat file will be automatically created. Then increase EXPECTED_VOTES to the desired value and reboot. The quorum disk votes will count, if the quorum disk is NOT mounted, but the QUORUM.DAT file can NOT be created, if the disk is not mounted.

 

Volker.

Homer Shoemaker
Frequent Advisor

Re: quorum connection lost and then reagained - normal?

Very good to know that releasing and re-acquring the quorum disk is normal behavior.  And re-assuring.

 

It's not like I'm going to be rebooting nodes all the time (the production server ran for almost three years straight, constantly in use, without rebooting before this project.  But the message that I dismounted the volume incorrectly, that's nothing to be concerned about?

 

I read that about the EXPECTED_VOTES parameter, but I forgot to do it the first time.  :(  But at least I got a little practise using the conversation boot mode.)

 

Thank you for your help, Sir.

 

 

 

 

Bob Blunt
Respected Contributor

Re: quorum connection lost and then reagained - normal?

The messages about clustering at system startup seem 100% normal for your config/setup.

 

If you're concerned about dismounting the Quorum disk you could add it to the "site specific shutdown" procedure.  It might not DO anything but it can't hurt to try.  Unless you've got a LOT of information stored on the quorum disk then you shouldn't have much trouble rebuilding.  You don't want to let the data on the disk get too stale so I'd recommend that you issue a $ SET VOLUME/REBUILD=FORCE ddcu: on your quorum disk.  This can be done after the system has booted and if there's a less active time that would be better.

 

bob