RHEL3 dynamically rescan scsi bus to mount cluster luns

[Gibbs, Simon]

I¹ve been looking through the QLA2x00 documentation (we¹re using QLA2340 FC
HBA¹s) and found this little snippet:

9.6 How To Force Rescan Of Devices
----------------------------------------------------------------
Starting from driver version v6.06.00b12 there is support for the mechanism
which allows the user to force the driver to do re-scan of the devices to
allow a new device to be added. This triggers the driver to initiate lun
discovery process.
To do this from the command line:

# echo "scsi-qlascan" > /proc/scsi/<driver-name>/<adapter-id>
(qlogic driver re-scans)

Where <adapter-id> is the instance number of the HBA.

Once that has been done the user can then force the scsi mid layer to do its
own scan and build the device table entry for the new device:

# echo "scsi add-single-device 0 1 2 3" >/proc/scsi/scsi
(scsi mid layer re-scans)

Where "0 1 2 3" is replaced by your "Host Channel Id Lun".

The scanning has to be done in the above mentioned order. First the driver
(qla2300/qla2200 driver etc) and then the Linux scsi mid layer (i.e. OS
scan) after.

I guess this is along the lines of what I need to do as well as an
additional command to update the PowerPath paths?
I¹ll have a go as soon as I have a minute ­ if anyone thinks I¹m heading
down the wrong track let me know.....



From: <Gibbs>, Simon <Simon.Gibbs at informa.com>
Date: Thu, 02 Jun 2005 10:33:35 +0100
To: <redhat-list at redhat.com>
Subject: Re: RHEL3 dynamically rescan scsi bus to mount cluster luns

Hi,

This may end up as a double post as my first attempt appears to be caught in
limbo ­ so my apologies now if that is the case.

I¹m currently in the process of setting up a 2 node cluster but have run
into a few problems.
The nodes are in different sites running RHEL3 AS and connected by fibre to
EMC Clarion arrays with PowerPath doing the multipathing - the cluster
software being used is Legato AAM.
The arrays are running MirrorView to replicate the data between each array
with one a primary and the other a secondary.

At the moment I can attach the luns to the first node and mount the ext3
filesystem, but when attempting to failover to the secondary node I¹m unable
to mount the filesystem ­ if the secondary node is rebooted it can then
mount the filesystem. I think this may be related to how MirrorView manages
the primary and secondary arrays. For example if the primary node is active
the secondary node is unable to access the lun.
So when failover occurs I don¹t believe the secondary node has the luns
loaded into the kernel, which means I need to somehow rescan the luns before
they can be mounted (I guess the equivalent command in Solaris would be
devfsadm).

Does anyone have any ideas whether this is possible and how I would go about
doing it?
Thanks for you time.

Simon



********************************************************************************
The information contained in this email message may be confidential. If you are not the intended recipient, any use, interference with, disclosure or copying of this material is unauthorised and prohibited. Although this message and any attachments are believed to be free of viruses, no responsibility is accepted by T&F Informa for any loss or damage arising in any way from receipt or use thereof.  Messages to and from the company are monitored for operational reasons and in accordance with lawful business practices. 
If you have received this message in error, please notify us by return and delete the message and any attachments.  Further enquiries/returns can be sent to postmaster at tfinforma.com