iSCSI Woes

Warning…technical language ahead!

One of my summer projects was a new server for one of my elementary buildings, with an iSCSI SAN to go along with it.  I started using iSCSI technology last year and just love it; it’s handy to be able to add space to a volume on the fly and the cost is substantially cheaper than fiber channel. The volumes that you create can be formatted any way you like, so it’s compatible with any file system; we are a Novell shop, so NSS is our standard.

The server setup was fairly straightforward after I discovered that Netware 6.5 SP6 was not going to fly with our new Dell hardware; the 6.5 SP7 overlay disk worked fine, however.  A little side rant here…Dell does not make enterprise quality servers.  Their pricing is good, but you get what you pay for; give me an IBM or HP any day.

With the server setup complete I added the volumes for staff and students, then migrated the home directories to it and rewrote the login scripts; everyone was happy and joyous.  Three days after the server was installed I got an NSS popup notification on my workstation announcing that one of the volumes had dismounted.  A quick look at the console confirmed this, so I issued the command to restart the iSCSI session and the volume mounted.  Weird. I opened the management console for the SAN and things looked okay with the exception of the occasional flicker of “failed” next to an iSCSI session.  That was bothersome, so it was time to hit the knowledgebase.

Every reference to iSCSI suggests that solid comms are in order; because of the way the protocol is designed there is little tolerance for duplex mismatches, flaky switches and bad cabling.  Our core switch is a big Cisco 6509 which is fairly new and a check of the ports did not turn up any duplex issues; the cables were new as well. After checking the most obvious things I began to compare this new setup with the SAN that had been happily humming along in my data center for over a year.  The Broadcom drivers that had been installed were out of date, so I upgraded them along with the latest winsock files from Novell.  A couple of days later the server dumped another volume.

I was starting to get really worried, because school was starting in a few weeks and people get annoyed when they are suddenly disconnected from their files.  A call to the SAN vendor was an exercise in futility because while the unit is Novell YES certified, Novell support is tough to get from most vendors.  That’s a rant for another day.  After another comparison of my known good server and SAN combination I noticed that the iSCSI drivers on the Novell server were old!  The latest version out was 1.06 and my brand new server set up with the latest service pack was at 1.03; how in the world did the old drivers manage to make it into the latest software download from Novell?

After digging around a bit, I found an iSCSI update on the Novell site and downloaded it. Part way through the install it quit with a message about only being written for Netware 6.5SP6, so it was apparent that I was not going to be able to use the scripted install routine.  After weighing my options and getting a good backup, I decided to replace all the files on my server with the files in the update package by hand which is definitely not for the faint of heart.  That meant unloading the iSCSI sessions, renaming all 19 of the files and replacing them with the new ones, then restarting the sessions. There was a moment of alarm when I realized that I needed to edit the newer ion.ncf file, but once that was done the volumes mounted properly and have been solid for several days.

I realized with some amusement that you would never get away with doing something like that in a Microsoft operating system; swapping files around by hand in Windows usually leaves you with a paperweight.  Netware can be a nuisance, but its modular nature makes it easier to deal with in some ways.

Let the school year begin!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: