Operating System - HP-UX
1835184 Members
2145 Online
110077 Solutions
New Discussion

Re: MC/SG newbie question

 
Gene L. Huft
Occasional Advisor

MC/SG newbie question

Gurus, it's been awhile since I've played on HP-UX...so I've got several questions - 1 at a time:
Scenario: I've got a 16 proc 8400 split up into 2 8 proc domains, it is apparently running with Clustered Volumes, using disk space from a SAN. The logical volumes are being used raw by Oracle - no filesystems are created.

I have a test whereby I just want to halt the cluster & use one node to build some actual filesystems on these raw logical vols. But when I do a cmhaltcl -f, I get an error message: "Clustered Volume Groups must be deactivated before the cluster can be halted."

1. So, my 1st question is How do I deactivate the Clustered Volume Groups?

2. Do I then have to use vgchange or something to tell the Volume Group (vg_ops) to NOT be cluster aware?

3. There doesn't seem to be any packages configured to start/stop when the cluster get's iniitiated...but is there anything else I'd need to do to the MC/SG config to tell it I don't want it to even start up upon reboot?


That's good for now, Thanks in advance!!

--Gene
6 REPLIES 6
Geoff Wild
Honored Contributor

Re: MC/SG newbie question

Can you post a cmviewcl -v

You don't need to halt the cluster to create filesystems...

To take existing ones out:

vgchange -c n /dev/vgXX

Rgds...Geoff
Proverbs 3:5,6 Trust in the Lord with all your heart and lean not on your own understanding; in all your ways acknowledge him, and he will make all your paths straight.
Gene L. Huft
Occasional Advisor

Re: MC/SG newbie question

HP-UX athprs01 B.11.11 U 9000/800 2774712589 unlimited-user license
# cmviewcl -v

CLUSTER STATUS
hp_rac1 up

NODE STATUS STATE
athprs01 up running

Network_Parameters:
INTERFACE STATUS PATH NAME
PRIMARY up 0/0/0/1/0 lan0
PRIMARY up 0/0/12/0/0 lan2
PRIMARY up 0/0/14/0/0 lan1

NODE STATUS STATE
athprs02 up running

Network_Parameters:
INTERFACE STATUS PATH NAME
PRIMARY up 1/0/0/1/0 lan0
PRIMARY up 1/0/12/0/0 lan1
PRIMARY up 1/0/14/0/0 lan2
#
Geoff Wild
Honored Contributor

Re: MC/SG newbie question

Try halting the nodes first - 1 at a time...

cmhaltnode athprs01

cmhatnode athprs02


The cmhaltcl -f should do that for you - sounds like maybe there is a volumegroup that is cluster aware - yet not in this cluster.

Check the vg's in the cluster ascii file...

Rgds...Geoff
Proverbs 3:5,6 Trust in the Lord with all your heart and lean not on your own understanding; in all your ways acknowledge him, and he will make all your paths straight.
Gene L. Huft
Occasional Advisor

Re: MC/SG newbie question

I'm still getting the 'must be deactivated' messages....where is the ascii cluster config file?

# cmhaltnode -f athprs01
Disabling package switching to all nodes being halted.
Warning: Do not modify or enable packages until the halt operation is completed.

Halting cluster services on node athprs01.
Unable to halt the LVM daemon on athprs01.
Clustered Volume Groups must be deactivated before the cluster can be halted.
Check the syslog file for more detailed information.

# cmhaltnode -f athprs02
Disabling package switching to all nodes being halted.
Warning: Do not modify or enable packages until the halt operation is completed.

Halting cluster services on node athprs02.
Unable to halt the LVM daemon on athprs02.
Clustered Volume Groups must be deactivated before the cluster can be halted.
Check the syslog file for more detailed information.
Geoff Wild
Honored Contributor

Re: MC/SG newbie question

THe cluster file should be in:

/etc/cmcluster

It may be called cluster.conf - but could be anything - depending on who cretaed it.

Better to grab it from the existing "live" binary:

cd /etc/cmcluster

cmquerycl -c hp_rac1 -n athprs01 -n athprs02 -C hp_rac1.ascii

To actually grab the lvm info (which will tell you which vg's are cluster aware):

cmquerycl -v -c hp_rac1 -l lvm

Rgds...Geoff
Proverbs 3:5,6 Trust in the Lord with all your heart and lean not on your own understanding; in all your ways acknowledge him, and he will make all your paths straight.
melvyn burnard
Honored Contributor

Re: MC/SG newbie question

Well this looks like you have no packages running which is a bit strange. I would assume you have a cluster lock disk, being a two node cluster./ If so, hten there may be a shared vg tha is activated.
I would suggest doing a cmviewconf, or run cmscancl, and get the info out of the cluster binary to see what is actually configured. If it does list a cluster lock VG, then do vgdisplay on each node and see if it is activated. If so, vgchange -a n
then try to halt the cluster.
If this is a newly created cluster that is being tested, then I can see how this might happen.
Also, are there any processes using any of the raw lvols in hte vg? if so, you would need to stop them first (normally done from within a package).


To stop it starting on reboot, change the 1 to 0 in /etc/rc.config.d/cmcluster
My house is the bank's, my money the wife's, But my opinions belong to me, not HP!