Link to home
Start Free TrialLog in
Avatar of nathanwilliams
nathanwilliams

asked on

ESXi 4.1 Tandberg LTO-4 tape drive dead path. Adaptec 29320LPE SCS Controller

we have a Dell PowerEdge 2950 using a tandberg LTO-4 HH with an Adaptec 29320LPE SCSI Sontroller

ESXi 4.0 works,    ESXi 4.1 does not see the naa.id of the tape drive and therefore treats the path to the drive as dead.

Any help would be appreciated if anyone has a solution to this.

We have an open case with VMWARE in which they are submitting this to the developers as a bug issue.  We have no resolution yet as of 8/9/2010
At this point we are forced to live with ESXi 4.0 until VMWARE fixes the issue.

Initially we were told by VMWARE support to contact Tandberg which we did.  Tandberg support said they do not test their drives on ESX.

VMWARE support then stated that they did not support tape drives and that there was only 1 listed in the HCL for ESX 3.5.

I explained at that point that our option was to switch to another Hypervisor such as Hyper-V or XEN if we could not get the drives to work as we do not have the budget to replace all of the tape drives at our remote locations, or to stand up physcial servers at each of those locations for backups.

Within 2 hours of that conversation, I recieved a call back asking for diags from esxi 4.0.0 and ESXi 4.1 and they would treat this as a bug issue.

As of yet we do not have a resolution and hope to have one soon.

In the mean time if anyone has a fix for this, that would be really helpful.

Thank you
Avatar of David
David
Flag of United States of America image

Tape drives are notoriously stressful and intrusive on a host machine. This is why you need to make sure they are put on a dedicated controller.  They generate bus resets under normal operating conditions among other things, and I'm sure you have experienced first-hand the evils of arbitration.

Just curious, are you running them on a dedicated controller?
Avatar of nathanwilliams
nathanwilliams

ASKER

The 29320LPE is dedicated just to the tape drive only and has performed great for us under esx 3.5 and esx 4.0 with Backup Exec 12.5       Starting with ESX 4.1 it no longers see the naa.id of the device.   It does however see the tape drive, SCI ID and LUN.  We have tried different scsi IDs and it recognizes the changes, just still not the naa.id so it treats the path to the device as dead.  
The 29320LPE is dedicated just to the tape drive only and has performed great for us under esx 3.5 and esx 4.0 with Backup Exec 12.5       Starting with ESX 4.1 it no longers see the naa.id of the device.   It does however see the tape drive, SCI ID and LUN.  We have tried different scsi IDs and it recognizes the changes, just still not the naa.id so it treats the path to the device as dead.  
That is very different, sometimes people report something "works", w/o beating it up in production.  In this case, it is clear this is a bug in ESX 4.1.  That call to tandberg made no sense, as the tape drive does not need any specific drivers beyond whatever your adapter card requires.  I've got programming info for many of the tandberg devices, not that one in particular, but I can't imagine the tape caring about any operating-specific settings when it comes to responding to the SCSI INQUIRY command or enumerating VPD pages.

Looks like you just got somebody in VMWARE support who wanted you to go away :(

There are no configurable tape-specific parameters that you could change within the tape, and while you could change a few settings on the HBA, there isn't anything that would affect the nature of the problem.  Looks like classic vmware programmer error.  Sorry, can't give you any advice beyond cross fingers. (Well there are some FC-> SCSI) bridges out there that let you attach tape via fibre channel interface, but they are expensive and no guarantee.  But if you are in a bind and want to try it, then there are enough changes in the architecture from SCSI to FC-attach, that it might be worth trying if you have the hardware and just want to get some backups done.

The first tech support person was useless.   The response was run esxcfg -mpath  and look we can the device, jus the path shows up as dead.   Contact the vendor as its not a VMWARE issue.
Called back asked for another tech and got teh Tape drives are not supported in VMWARE.  read my initial post.   Very frustrating.     We have come older HP drives that work just fine with ESX 4.1, but they are DLT drives and we have switched over to LTO-4.  I noticed there are other users in the same boat as I am with Tandberg and IBM single tape LTO-4 drives.
Well, the LTO-4s are expensive investments, so you are at mercy of VMWARE to investigate.  Personally, I would not count on a solution, seeing as you have identified other users in same boat.  VMware 4.0, or .. there are some low-cost FC-SCSI bridges out there, it may be worth trying, found this in ebay .. but buy at your own risk, and no guarantee.  Maybe if you get get somebody in VMWARE Level2 to help you understand the exact nature of the bug, like what is different between 4.0 and 4.1 discovery, that will give you a clue.

Here is link to something I found in ebay (at your own risk, this is just a shot in the dark)
http://cgi.ebay.com/Adic-SNC-3000-1GB-2-FC-x-4-LVD-68p-SCSI-Fiber-Bridge-/180401639750?pt=LH_DefaultDomain_0
ASKER CERTIFIED SOLUTION
Avatar of reitzner
reitzner
Flag of Germany image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Great Reitzner ... the command makes sense, and easy enough to test.  It comes down to whether or not the LTO4 presents LUNs.  While you can figure this out with diagnostic software, it is so much easier just to try the command.

Make sure that you tell SCSI adapter to NOT use LUNs / logical addresses, or do LUN scanning.  (Can't remember if it is an option, but you want feature disabled if it is)
Will try this today and provide feedback.
Thank you
Will try this today and provide feedback.
Thank you
Two great articles that go into detail explaining the pocible solution of esxcli nmp satp deleterule --satp VMW_SATP_ALUA --claim-option tpgs_on    provided by Reitzner

http://www.yellow-bricks.com/2009/03/19/pluggable-storage-architecture-exploring-the-next-version-of-esxvcenter/

http://blogs.vmware.com/storage/2009/10/vstorage-multi-paths-options-in-vsphere.html

I will post whether it works or not later today
Worked like a champ
Thank you everyone for the quick responses on this.

This is the command that fixed the issue.
esxcli nmp satp deleterule --satp VMW_SATP_ALUA --claim-option tpgs_on
This is the way that we finally fixed it after support got back to us from VMWARE (Only took 2 weeks)
Tape Drive support in ESXi 4.1 using adaptec 29320LPE Controller
In VCenter or VSphere client verify that the tape drive path under storage adapters shows a dead path to the tape drive and or changer

To fix the dead path issue
Enable local and remote support via the ESXi 4.1 console
Alt F1
Login as root
Run the following command
esxcli nmp satp addrule --satp VMW_SATP_LOCAL --driver="aic79xx" --description="Specific rule for Adaptec Card"
Reboot the ESX server
login to vcenter or vsphere client and you should now see your SCSI tape drives

IMPORTANT NOTE:
If you are using a tape changer
In the Adaptec BIOS, you need to change the option to enable multiple LUN support for the SCSI ID of the changer in order for windows to recognize the tape drive and changer properly.  When you reboot and the adapter controller initializes instead of seeing 1 device, it should show 2 the tape drive and the changer.
esxcli nmp satp addrule --satp VMW_SATP_LOCAL --driver="aic79xx" --description="Specific rule for Adaptec Card"
The command above will also work with other scsi drivers.   just replace the aic79xx with the name of the driver loaded for your scsi controller and it wil probably work.    We had success with an HP scsi controller with an LSI logic chipset that worked well too.