Re: HACMP remove vpath/hdisks on standby server
From: Charles THISE (charles.thise_at_gb.be)
Date: 07/04/03
- Next message: Alberto 'JCN-9000' Varesio: "Re: Ping packets from AIX"
- Previous message: Xolkz: "Re: Rebooting HMC and p650"
- In reply to: Jeff Wilson: "Re: HACMP remove vpath/hdisks on standby server"
- Next in thread: 4nimal: "Re: HACMP remove vpath/hdisks on standby server"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: 3 Jul 2003 23:40:28 -0700
jeff.wilson@gwl.com (Jeff Wilson) wrote in message news:<861eec5a.0307021122.1fb20732@posting.google.com>...
> uk03170 <member30505@dbforums.com> wrote in message news:<3061703.1057062006@dbforums.com>...
> > Hi, I need some advice regarding removing the a pv on a standby server.
> > Both servers are running HACMP 4.4.1, AIX 4.3.3 (ML 9), DLM.
> > The volume group on the primary server consists of 3 vpaths with 2
> > logical volumes, I am going to remove one of the logical volumes which
> > spans 2 of the disks (vpaths) in this volume group, this will free up
> > one of the vpath devices completely which means it can be removed from
> > the volume group with a reducevg and then I can rmdev the vpath and
> > associated hdisks, thats the easy bit.
> > The problem I then have is with the standby server as it still has the
> > vpath/hdisks associated with the volume group that has been reduced on
> > the primary server, to the best of my knowledge CSPOC will not work with
> > vpath devices and the only "safe" way of doing this is to vary on the
> > volume group on the standby server and export and import the volume
> > group with HACMP closed down, I can't close HACMP as I cannot get down
> > time to do this and need to do it whilst HACMP is running, I can't rmdev
> > the devices as they are associated with a volume group and I don't
> > really want to use odmdelete, a colleague suggested that I leave the
> > standby server as it is and take away the Lun at a fabric level so the
> > next time the server is rebooted the devices will be in a defined state
> > and we can then remove them, I'm not keen on this idea as it's untidy
> > and I don't know when we will be able to reboot this server.
> > Thanx
>
> I have had the same problem. I am currently working with IBM to fix
> this problem but the only recommended way to hadel this is to bring HA
> down and do the export import. One note if you have already removed
> the disk from your primary server and not on your stand by your
> enviroment is out of sync and could halt the machines.
You can solve your situation by these operations :
1. on the primary : unlock your vg by varyonvg -n -b -u <vgname>
2. on the standby : import your vg on an existing pv by imortvg -L
<vgname> <pvname>.
This operation update the ODM on standby machine by reading VGDA/VGSA
on each disks from this VG.
3. on the primary : lock your vg by varyonvg <vgname>.
Warning ! this kind of operation is really dangerous. Keep in mind in
wrong case, you must put varyoff this vg whithout any stop of HACMP.
Hope that can help you.
Good luck.
Charles.
- Next message: Alberto 'JCN-9000' Varesio: "Re: Ping packets from AIX"
- Previous message: Xolkz: "Re: Rebooting HMC and p650"
- In reply to: Jeff Wilson: "Re: HACMP remove vpath/hdisks on standby server"
- Next in thread: 4nimal: "Re: HACMP remove vpath/hdisks on standby server"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
|