Hi, scenario like below: currently i have 14 lovls in same pv,vg. runnign database and applicaton.now with the additional hdd i will create new pv with the same vg.then i will mirror all 14 lvols to the new pv.and remove lvols from old pv. i like to run application and database from new pv.should i have different mount point or it will run from separate mount point? this is under HP UX SG cluster. plz help.
Hi, understood. but my question is once i reduce the existing lvols and like to keep mirrored lvols. should i create different mount point? or it will run from existing mount point.
You are mirroring your lvs and that doesn't need a different mountpoint. Mirroring actually duplicates the extends to the new pv and the filesystem not aware of the underlying redundancy level (mirror) as the lvm manages the mirroring.
The vgimport command adds the specified volume group to the system. The physical volumes, specified as pv_path ..., are scanned to obtain the volume group information and logical volume information. This command works much like vgcreate by requiring that the volume group device directory and the group special file be created before the command is executed (see vgcreate(1M)). The vg_name is added from the /etc/lvmtab file, and the associated logical volume device files are added to the system.
As you are going to remove your PV by mirroring it to new and detaching the old, you need to replicate the changes to other cluster nodes by exporting vg in preview mode to a create a map file and import it on other cluster node.
Recommend to create a runbook for your activity including all the steps to complete an activity. You can probably keep a latest backup as a precautionary measure and execute the task during hours with less load etc.
Yes, After completing everything on node1 you can create map file and copy it to other node.
You can see this thread having the steps to export and import below, but recommend you to create your own runbook with all the commands you are going to execute in the right order and verify and ensure it is fine. This will help to avoid forgetting/missing something to do.
what you are trying to do is called "host based mirroring". Basically, it goes online, without even stopping the server or the application, but for this you should have the product MirrorDisk/UX installed. You should migrate it on the active server (the server your applications are running currently), after that you should bring the changes also to the passive node. (vgexport -p on the active node, the import the mapfile on the passive node.) No mountpoints/cluster config should be changed.
Refer the attached LVM docs from HP. This document discusses how to deal with disk failures under the HP-UX logical volume manager and it includes strategies to prepare for disk failure, ways to recongnize that a disk has failed and steps to remove and replace a failed disk.
