how the assignment of new Virtual Disk to the Bay appears without reboot Linux

[Maaz]

Dear Gurus,

Bay 1 is running SUSE Linux Enterprise Server 10 SP 1

Using HP EVA System Management Server, when we assign a new Virtual Disk to the Bay 1(SLES 10 SP 1), the new disk(SAN device) doesnt available/appear to the OS(fdisk, or yast) unless we reboot the OS(SLES 10).

or

if we extend/increase the size of an existing Virtual Disk, the increment in the size of virtal disk doesnt appear, unless we reboot the system


is it possible that without reboot the OS, the new virtual disk or the incerement in the size of an existing virtual disk will be available to the OS(by restarting any service)

Regards

[Ivan Ferreira]

If fibreutils package is available for SUSE (check the download page), you can use the hp_rescan command to recognize the new luns. You can also try the rescan-scsi-bus.sh script (search it in google).

For the partition size, you may try with the partprobe command.

[Ross Minkov]

Maaz,

Yes, the hp fibrechannel utilities RPM (fibreutils) is available for SLES. It's part of the ProLiant Support Pack (PSP). Probably a good idea to install the whole PSP, not just parts of it. Once you have fibreutils package installed run hp_rescan and then lssd/lssg.

HTH,
Ross

[Maaz]

Thanks Ivan Ferreira, and Ross Minkov for nice help.

To recognize the new Virtual Disk(LUN), following commands works

# hp_rescan -a
# lssd
# lssg

DONE.

But still we have to reboot the SLES 10 SP 1, to recognize the increment in the size of an existing VDisk.

To recognize the increment in the size of an existing VDisk, after increasing the size from EVA SMS(Windows 2003 R2 Storage Edition), i tried following commands
# partprobe
# hp_rescan -a
# lssd
# lssg
but still YaST, and fdisk shows the old size.

But Once i reboot the OS, YaST/fdisk reflects/shows the increment(I have to resize the filesystem).

Once again Thanks for nice and superb help.
Regards

[Maaz]

Ivan Ferreira, instead of Virtual Disk, you wrote LUN... why ?

one more thing to ask.

Once we assign/present the VDisk to the wwn port of our two machines(SLES 10) should i have to configure the disk(partition, format, mountpoint) on the both the Servers(machineA, and machineB)? Or just configure the Disk(partition, format, mountpoint) on a single machine, and it will automatically reflect on the other machine ?

Likewise if i create a file on that SAN Disk(VDisk) from machineA, does the same file will also be accessable from machineB ?

Regards

[Ross Minkov]

Maaz,

LUN vs. Vdisk: LUN is Logical Unit Number -- an address for an individual disk drive and by extension, the disk device itself. The term is used in the SCSI protocol as a way to differentiate individual disk drives within a common SCSI target device like a disk array. Vdisk is the EVA terminology. When you talk about the EVA use Vdisk, when you are on a host you see LUNs.

Re: re-sizing filesystems -- do you use LVM?

-Ross

[Ivan Ferreira]

>>> Once we assign/present the VDisk to the wwn port of our two machines(SLES 10) should i have to configure the disk(partition, format, mountpoint) on the both the Servers(machineA, and machineB)? Or just configure the Disk(partition, format, mountpoint) on a single machine, and it will automatically reflect on the other machine ?

You only need to do it on one machine

>>> Likewise if i create a file on that SAN Disk(VDisk) from machineA, does the same file will also be accessable from machineB ?

No unless you use a cluster file system like Red Hat GFS or Oracle OCFS2. Without that, don't even try to mount the file system on both machines as you will cause data corruption.

See also the attached file.

[Maaz]

>You only need to do it on one machine
If I configure the disk on machineA(and all the configuration will be available on the other machineB also)then it means that on both machines the disk will be mounted automatically, but Ivan you said that:

>don't even try to mount the file system on both machines


In case I dont mount the disk on the other server(machineB), and also didnt configure the cluster Filesystem(OCFS, GFS). Then if the machineA is down(properly shutdown), and I mount the VDisk on machineB, will I get the complete data wihtout any issue ?

I am asking again and again, Please dont mind, I dont want any mishap/mistake, thats why I am asking everything from you Gurus. Any minor detail/info is very much important for me.

Regards
Maaz

[Ross Minkov · Accepted solution]

Maaz,

It looks to me like you are doing something like a cold standby server -- in the event you have a hardware problem with the primary for a longer period of time, you want to bring up the standby server and have it do the work that the primary was doing. In this case I would try to configure the standby server the same as the primary and (assuming you have your data on that SAN disk in question) I wouldn't even present it to the standby from the Selective Storage Presentation (SSP) on the EVA. If/when the primary server goes down you would go to the EVA and change the Vdisk presentation, then rescan (like we talked before -- btw, make sure you have the PSP installed on both servers) from the standby and then mount the data disk on it.

I might be off on what you are doing there, but do let us know.

HTH,
Ross

[Ivan Ferreira]

>>> In case I dont mount the disk on the other server(machineB), and also didnt configure the cluster Filesystem(OCFS, GFS). Then if the machineA is down(properly shutdown), and I mount the VDisk on machineB, will I get the complete data wihtout any issue ?

Yes, you will access without any issue as long as the other server does not starts. So, removing disk presentation from the other host would be mandatory, this can be done from the storage or using sanswitch zones if you use sanswitches.

A better solution for this would be a software like Red Hat Cluster Suite or ServiceGuard. But in this case, SLES maybe should be replaced =(