Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations Mike Lewis on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

HBA/Controller problem in Solaris 8

Status
Not open for further replies.

MorganGreylock

Programmer
Jan 30, 2001
223
US
We have a couple servers running Solaris 8, connected to an EMC SAN frame. There are two connections (two HBA ports) to allow DMP for redundancy. Sun has suggested some patches, but none of them have worked. The SAN team sees that both HBAs are logging into the EMC frame, but format still only shows a single controller. (It should show c2 and c3, but only shows c2).

Various commands and whatnot haven't helped, and EMC is suggesting that we replace the HBAs. We just think it's odd that it happened to two servers that were built at the same time, sit in the same rack, connected to the same switch.

Anyone have any ideas about what else we can try before we replace the HBAs?

Thanks in advance!
 
What type of HBAs are they? There may be all sorts of funky things you need to do in /kernel/drv/sd.conf and /kernel/drv/drivername.conf to make sure that both HBAs are logging in to the fabric correctly and seeing the correct ports, etc.

Samples of the contents of those files, especially details of any customisations you have made, would help us troubleshoot. It sounds unlikely that you would have two identically faulty HBAs in different servers...

Annihilannic.
 
The two servers have QLogic cards, which don't use the sd.conf file. Some of our older boxes have emulex cards that use lpfc.conf and sd.conf, but these were replaced some time ago with the qlc/leadville drivers, that don't require those.

I agree completely that it is unlikely that both had a failure at the same time, but since the date/time of the original problems were not known, it could have been a situation where one failed, then a year or two later the other server failed. They were identified with a by-request scan by our storage team because of a frame update that they will be performing.

Thanks
 
I would be very surprised if the QLogic cards don't rely on sd.conf since (to my knowledge) that is the driver used by any SCSI disks on a Sun system. If you don't configure enough SCSI 'slots' in sd.conf, no matter what you do at the vendor driver level you will not see all of your disks.

Annihilannic.
 
Qlogics with the Leadville driver don't use sd.conf... at least that's what I've been told. The file exists but doesn't have hardly any lines it at all. We have hundreds of them in production already. The issue here is that something caused one of the controllers to disappear from the system, but they used to be there at some point. The servers would have never been allowed to be put into production without both controllers active, that's one of the things we check when they are built.

I'll still double check this at the office tomorrow... never know!

Thanks for the responses, give me all you have, we're desperate over here!

Thanks!
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top