VG on EVA Dismissed After system reboot

[teckmofyc]

Hi everybody, I have 2 rx3600 server with hpux 11.23 and an EVA array of 1Tb who is configured on mirror in my two systems. It appears as a one Physical Volume of 512Gb by every server.

I can see the PV by sam and I can create a VG for every server looking the same PV creating the VG>LV>And FS. But at the moment of rebooting the two systems, One of them loses the the VG configuration the VG give a error as it were inactive and. After that when I enter on SAM and I try to create again the VG and everythingelse I start deleting the VG and all its association, this error appears.

The command used to remove a logical volume, /sbin/lvremove, has failed. The stderr output fromthe command is shown below. logical volume has not been removed.

thanks in advance and easy points to who can helpme to solve my problem.

[Duncan Edmonstone · Accepted solution]

Hi,

>> I can see the PV by sam and I can create a VG for every server looking the same PV creating the VG>LV>And FS

Are you saying that you create a set of vdisks on the EVA, and then present the same set of vdisks to both your hosts with the intention of using theem to see the same data on both hosts? (i.e. vdisk1 is presented to hostA and hostB and you expect to see the same VG/LV/filesystem on both hosts)

If that's what you mean, you can't do that unless you have either Shared LVM (SLVM) or Symatec/Veritas Cluster Volume Manager/Cluster Filesystem (CVM/CFS) - both of which imply you are trying to use Oracle RAC. For any other purpose, your app may not be able to manage concurrent access from 2 hosts.

Take a step back and tell us what you want to acheieve here and we'll tell you if that is possible with the tools you have...

[teckmofyc]

Hi, in effect this big volmune is created with the purpose of been seen the same data from both host A and B as you said, but there is one detail that my boss told me, that maybe will make you get rid with the answer that I need, In the servers, are supose to have installed Service Guard, and being in clusters. Does serviceguard would manage the configuration of the Volume Group and logical volumes, and I just need to present them the physicals volumens=the disk as comes y tell to person that gonna install serviceguard that he configure the volumes? Or there is another solution for me?

[Tim Nelson]

To your last point YES> You cannot have more than one server access the same device at the same time unless it is managed by a clustering application like SG, even in this scenario it is an active/passive mode.

In order to do active/active in addition to cluster software your application must also be active/active capable ( there are few of these around.

[Duncan Edmonstone]

If the volumes are going to be used for Serviceguard I'd advise you to completely remove the configuration from both hosts, and then just re-create it on one host. When the Serviceguard conultant arrives, just point him at the filesystems and volume group on that one host that you want clustering. He/She will know what to do with it from that point.

HTH

Duncan

[teckmofyc]

thanks guys yes the answer es installa first the service guard, and make it create de volumens or export the volume and import the volume as read only permisology from the second server and then create the VGs. Thanks to All!