vxdiskadm gives: ERROR V-5-2-1207 vxconfigd has to be running

[khemmerl]

Hello everyone,

I'm in a bit of a sticky situation in that I have to get SFHS working for an Oracle 10g DB server within 2 weeks with no training.  To be honest I'm not sure if this problem should be posted under Storage Foundation or Cluster Service as I don't understand what the difference is. 

So far, I've been going through the Veritas Cluster Server Installation Guide (Solaris) 5.1 and have successfully completed everything in the pre-installation and installation chapters.  I am now on chapter 7 Configuring VCS clusters for data integrity and am attempting to set up disk-based IO fencing.  Following the instructions, I have run devfsadm on both servers - this command did not return anything.  Next I ran vxdiskadm and selected option 18) Allow multipathing/Unsuppress devices from VxVM's view but it immediately gives the following error:

 Select an operation to perform: 18

Include Devices
Menu: VolumeManager/Disk/IncludeDevices
  VxVM  ERROR V-5-2-1207
The Volume Manager Configuration Daemon (vxconfigd) has to be running in
  order to continue this operation. Please exit this program, start vxconfigd
  and proceed with this operation.

Hit RETURN to continue.

I can't find any information on the error code V-5-2-1207 and searches for "vxconfigd not running" haven't found anything that I think is applicable. 

Any suggestions on what I missed would be greatly appreciated.

Thanks

Ken

[khemmerl]

I tried a "vxconfigd -m enable" but got the following error:

 V-5-1-11887 vxvm:vxconfigd: NOTICE: Generating /etc/vx/array.info

VxVM vxconfigd ERROR V-5-1-1589 enable failed: Volboot file not loaded
        transactions are disabled. 

Ken

[mikebounds · Accepted solution]

It sounds like you have not configured volume manager - to do this run "vxinstall" - this is a menu driven interface.

I believe the defaults for the questions asked are:

Use Enclosure-Based Naming

Set-up a System-Wide Default Disk Group

But in my experience most customers do not use Enclosure-Based Naming (they use O/S naming scheme) and do not set-up a default diskgroup, but you can do if you are only going to create one diskgroup.

You can change naming scheme using vxdiskadm at anytime, so it doesn't really matter which one you choose.

This should have been posted under SF as SF is Volume Manager + Filesystem (i.e the Storage) where as cluster is about making servers redundant.

Mike

[khemmerl]

Holly crap, vxinstall started up a whole bunch of stuff that I'm not sure I'm licensed for:  
vxrelocd, vxnotify, vxconfigbackupd, vxattachd, vxesd, vxconfigd, vxconfigbackupd, vxattachd, vxdbd, vxnotify, and vxrelocd.

I've tried rebooting but everything is still there.  Hopefully I haven't just screwed my install.

[khemmerl]

Some good news: "vxdctl mode" now reports "mode: enabled".  I'll continue on with my configuration and trust that all those processes are supposed to be running.

[mikebounds]

You only need a VxVM licence (part of SFHA) for all these processes (i.e you should have an SFHA licence). If you want to use I/O fencing, then you need VxVM.  If you don't use I/O fencing (this is only mandatory for CFS and RAC - i.e accessing storage concurrenty from more than one server), then you don't HAVE to use VxVM & VxFs and you can use a combination of ZFS, SVM, ufs or Solaris slices instead.

vxfs will gives you better performance than ufs for Oracle data files.

Mike

[khemmerl]

Thanks for your help Mike.  After running vxinstall on both nodes I'm now able to continue on with setting up disk fencing.  I've marked your post as the solution.

Ken