Connecting SAN storage to NW6.5

[Datong]

Hi All,
I have a test server that I am trying to get to connect to our SAN. The SAN has been setup so that I have a 20GB test area to play with. The Zones and LUN have bee setup and I have an HP (Emulex P1150) HBA card installed and working. I have set the multi=path support=on so that I only see one device. I can initialize the device and partition, create the pool but not the volume. Also if I reboot the server the settings are not saved. I have tried adding scan all luns to the autoexec.ncf before the mount all. I also had a warning when activating the pool that my clustering software was not running. Im not clustering the server so is this cosmteic or do I need some clustering component to be running ?

NW6.5 SP7
HP Badged Emulex P1150 HBA (1 firbre card 2 channel)
HP EVA4000 SAN.


autoexec.ncf sample

SET MULTI-PATH SUPPORT=ON
SCAN ALL LUNS
MOUNT ALL

Any advice much appreciated.

 

[sstoppel]

So you can create the pool but not the volume? I've never seen that before.

FYI mount all will mount VOLUMES, so if you have no volume defined mount commands won't do anything for you.

 

[marvhuffaker]

I've also never seen a case where you create a pool but couldn't create volumes. That's very unusual. What error messages do you get when you try to create the volumes? Typically, if the LUN is presented to the server, you can create the partition/pool/volume no problem at all. You could however try to Initialize the device again (in NSSMU on the Partition menu). When you create a volume, it also creates a volume object. Sometimes you get errors there, but the fact that you create the pool (which also creates NDS objects) makes me think NDS and the overall function of NSS is working.

Also, I think you are going wrong with your statements because the HAM drivers load in your STARTUP.NCF, but you have those commands in your Autoexec. Try moving them to your STARTUP.NCF at the top.

Also, in some cases, when you load HAM drivers, you need to specify the LUN setting as well there. The syntax varies by device and I can't remember any specifics offhand. Sometimes it is simply LOAD xxx.HAM /LUNS but could be other. You need to check with the driver docs.

You also do not need clustering running. I'm not sure why you're getting that, it sounds like it thinks it has the shared flag enabled, but that is unlikely if its a new pool. Is there already data in this LUN?


Marvin Huffaker, MCNE
Marvin Huffaker Consulting, Inc.
A Novell Platinum Partner

http://www.redjuju.com

 

[Datong]

Hi all,
thanks for the replies. I was seeing two different errors initally, a 21711 on the pool which refused to go active unless I had clustering software running. Then when I managed to get the pool activated by recreating it I got a 606 when trying to add the volume to NDS. I think this was an error on my part as there may have been a volume object in NDS already from a previous attempt. The clustering error threw me though as Im not clustering the server nor will it be in production. What I found was that whilst I was playing around with settings and things I inadvertantly marked the device as Sharable for Clustering (as you mentioned in your reply). I removed this and reinitialised the device and started again. This time I saw no errors and I was able to create the partition, pool and volume successfully.
Marv - I looked at the drivers for the HBA card and there are settings but they cant be changed as they seem to be read only once it is running. You have to delete the card from the hardware list and reinstall it to make any changes. However I couldnt see or find any docs on the switches you mentioned although I have seen them in the SAN docs. I dont think I mentioned my HAM drivers but what did you mean about putting these commands into my startup.ncf ? I cant see any in my autoexec.ncf.
The drivers for the HBA load ok, it was just getting the volume to mount and re-mount when I restarted the server. The lines I added to the autoexec.ncf are :

SET MULTI-PATH SUPPORT = ON
ETC
ETC
INITSYS.NCF
SCAN ALL LUNS
?Y
MOUNT ALL

The multipath statement helped as it grouped the devices into one (fibre card has 2 channels). I let the INITSYS run to setup the network, then issued the SCAN ALL LUNS command which is required to get NW to see the device. I dont know if this is all correct but it seems to work in this instance. I put in a small delay ?y (default 10 seconds) for the scan to complete then just let the autoexec continue to mount all the drives and that appears to work.
I think the root cause of all my initial problems was marking the drive as shared for clustering, thus Novell was expecting clustering to be running which it isn't.

Thanks very much for your advice guys, as usual you have pointed me in the right direction and putting it all together seems to have solved this. I will continue my testing now to see if I can break it before going live with it. Any other advanced warnings/gotchas much appreciated.
Cheers.

 

[marvhuffaker]

Your ham drivers load from C:\NWSERVER\STARTUP.NCF, not the AUTOEXEC.NCF file. That is where you need the commands that control the driver when it loads. I don't think you want "Set Multi-path support=on" and "Scan All LUns" in your Autoexec, that gets loaded WAY after the initial driver loads when you actually need them. Put them in your Startup.ncf file.

You don't need to add/remove to change anything. You can unload the driver and reload it with the command line parameters you need.. I've never seen a driver that forced me to do anything otherwise.

Marvin Huffaker, MCNE
Marvin Huffaker Consulting, Inc.
A Novell Platinum Partner

http://www.redjuju.com