Windows Admin needs Linux help- Red Hat 4 hangs on reboot when LUN presented

[raylittlefield]

Hello,
Before I begin, I want to let you know I am not competent in Linux administration; but can follow directions pretty well :)

we have had a Proliant Dl580 G3 box with a single FCA2214 HBA running Red hat 4, and connected to an EVA5000 for several years (EVA running VCS 4.007)

Not knowing how to configure multi-path, we zoned only one port of the controller to the HBA, presented disks and everything was just great. We found if we zoned in the other controller, the OS and the Qlogic card could auto-magically figure multi-pathing out.

We recently purchased an new EVA 6400, and wanted to migrate the data from the EVA5000 to the 6400. So I created a new zone- one port of one 6400 controller to the fca2214 and presented a new LUN.

I ran the command 'HP_rescan -a ' and the system picked up the new EVA 6400 LUN. after partitioning and formating and such, I copied the data from the EVA 5000 LUN to the EVA 6400 lun and everything was great- until I rebooted.

The system hangs during boot right after mounting the local disk in read-write mode (sorry don't have verbatim message with me) there is no error message or additional information I can gather from the /var/messages logs. It just hangs. I found that if I unpresented the disk in command view- the server 'unhangs' and will continue to boot up and work fine.
I checked host properites in Command View of EVA6400 and it is set to Linux. Thinking maybe I formatted or did something wrong to the LUN, I have presented fresh LUNs to the box, but it always hangs when any LUN is presented from the EVA6400 upon reboot. I even tried removing the zone from the EVA5000 and just presenting the EVA6400- but it continues to hang on reboot with EVA 6400 LUN presented.

I updated drivers to the latest 8.02.23 but the symptom remains. Any suggestions?

[Rob_69_1]

Hi, is your system booting from SAN? And does it have a multipathed root device?

If it does, pls check the following link, it might be of help:

http://forums.oracle.com/forums/thread.jspa?threadID=946143&start=0&tstart=0

Basically, you need to upgrade the mkinitrd package on your system, and rebuild the initrd image that is going to be used at boot time.

RedHat 4 needs a newer mkinitrd to support multipathed root partition in a SAN environment, and to manage the fact that if you present new LUNs to the system, at the next boot the order of discovery of the devices (and their names) change.

HTH

[AnthonySN]

Maybe you can try updating the PSP

http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareIndex.jsp?lang=en&cc=us&prodNameId=3288118&prodTypeId=15351&prodSeriesId=408421&swLang=8&taskId=135&swEnvOID=2025#78221

btw what is the update version of Linux.

[raylittlefield]

Just wanted to update this thread with what I beleive was the problem and solution.

I loaded SANSurfer and went thru the settings of the Qlogic HBA card. One of the NVRAM settings is the connection type.

It was set to Loop Preferred, otherwise Point to Point. Not sure if that is default, but I set it to Point to Point Only, as that seemed appropriate for a fabric connection.

Anyways, that seems to have fixed it- I can boot with LUNS from both storage controller now without hanging.

Also here is a good article regarding multipathing in Redhat that I'm currently trying to configure:

http://www.redhat.com/docs/manuals/csgfs/browse/4.6/DM_Multipath/index.html