VMware Cloud Community
Jan_MS
Contributor
Contributor

ESX Ranger + VCB: "Too many SCSI reservation conflicts" error

Hello,

we use ESX Ranger + VCB to backup our VMs - only one backup simultanous![/b]

Most of the jobs are running without errors, but sometimes we have problems and get this message:

Backup Result: Error[/b]

VM Name: HP-PCM

VM Current Host: isc-esx-s03.isc.de

Log: Modifying VM Configuration For Backup.

Executing esxRanger(VCB) Process.

Status: Preparing to Launch VCB.

Status: Vizioncore VCB Backup Started.

Status: Plugin v. 1.0.3 Built on 02/09/2007

Status: Copying "\[Data170] HP-PCM/HP-PCM.vmx": 0%=====================50%=====================100%

Status: 170] HP-PCM/HP-PCM.vmx": 0%=====================50%=====================100%

Status: Copying "\[Data170] HP-PCM/HP-PCM.nvram": 0%=====================50%=====================100%

Status: 170] HP-PCM/HP-PCM.nvram": 0%=====================50%=====================100%

Status: Copying "\[Data170] HP-PCM//vmware-1.log": 0%=====================50%=====================100%

Status: 170] HP-PCM//vmware-1.log": 0%=====================50%=====================100%

Status: Copying "\[Data170] HP-PCM//vmware-2.log": 0%=====================50%=====================100%

Status: 170] HP-PCM//vmware-2.log": 0%=====================50%=====================100%

Status: Copying "\[Data170] HP-PCM//vmware.log": 0%=====================50%=====================100%

Status: 170] HP-PCM//vmware.log": 0%=====================50%=====================100%

Status: [2007-07-12 21:45:41.454 '

Status: App

Status: ' 5712 warning]

Status: Could not obtain inquiry page 128 for device on path 2, target 0, lun 0

Status: [2007-07-12 21:45:41.454 '

Status: App

Status: ' 5712 warning]

Status: Sending SCSI inquiry failed: Unknown error. (No proper error code was returned.)

Status: [2007-07-12 21:45:43.907 '

Status: BaseLibs

Status: ' 4568 warning]

Status: \[Vmdb_Unset] Unsetting unknown path: /vmomi/

Status: Converting "D:\.\HP-PCM\scsi0-0-0-HP-PCM.vmdk" (flat file): 0%=====================50%=====================100%

Status: [2007-07-12 21:46:53.205 '

Status: App

Status: ' 5712 warning[/b]]

Status: SanMpAIOMgrRWv: Too many SCSI reservation conflicts[/b].

Status: [2007-07-12 21:46:54.517 '

Status: App

Status: ' 5712 warning[/b]]

Status: CheckActive: Too many SCSI reservation conflicts[/b].

Status: [2007-07-12 21:46:55.517 '

Modifying VM Configuration For Backup.

Cleaning Up After Backup.

Determining Backup Statistics.

Writing Backup Information.

Writing Backup Information to the Database.

Any ideas to solve the problem? Clues, hints ?

Thanks!

Regards,

Jan

Reply
0 Kudos
8 Replies
Texiwill
Leadership
Leadership

Hello,

This is actually not an esxRanger issue but a problem with VCB. In effect VCB is causing many SCSI-2 Reservations within the SAN fabric which causes scsi reservation conflicts.

You will get the same results if you use vcbMounter direct from a cmd.exe shell.

Do to this, we decided to not use VCB. The number of locks VCB/Windows needs to set to do its stuff requires a very powerful (cache and performance) remote storage to support.

I found no solution. There maybe something you can do on the Windows VCB Proxy server but there is nothing that can be done from the ESX side, or within VCB.

Best regards,

Edward

--
Edward L. Haletky
vExpert XIV: 2009-2023,
VMTN Community Moderator
vSphere Upgrade Saga: https://www.astroarch.com/blogs
GitHub Repo: https://github.com/Texiwill
Schorschi
Expert
Expert

After walking down this road with VMware, we determined that this issue can be the result of incorrect Zoning in your shared-storage setup, and/or pathing software on the VCB proxy server. Every environment can be different so you need to look at both issues. For example, HP MSA1000, must be explicitly zoned, HP SSP presentation is not sufficient. And, Emulex HBAs load a new filter driver Emulex Plus, that works with HBAnywhere, but in fact, like pathing software, seems to create issues. Removing this filter driver, and explicit zoning, and we successfully tested up to 6 concurrent VCB snapshots on a HP MSA 1000, now of course you would not do this in real life, the performance is just not there on an HP MSA 1000 to support it. But it took 6 or 7 snaps concurrent, to generate any SCSI reservation errors, where before these changes, we had reservation issues with just 2 snaps concurrent.

Reply
0 Kudos
nitinsahi
Enthusiast
Enthusiast

Well.... VCB is not a complete backup solution..... Self Vmware adminits it has sevrals issue that need tobe fixed .... I think we need to wait for another 6 months till the time VCB come up with the zero error ..... There are other lockup issues like this also with VCB .....

Nitin

Reply
0 Kudos
Jwoods
Expert
Expert

For example, HP MSA1000, must be

explicitly zoned, HP SSP presentation is not

sufficient.

Removing this filter driver, and explicit

zoning, and we successfully tested up to 6 concurrent

VCB snapshots on a HP MSA 1000

What do you mean by "explicitly zoned?"[/b] Is this the same as LUN masking?

Reply
0 Kudos
BUGCHK
Commander
Commander

LUN masking[/b] is done on the storage array. It is a mapping of storage containers (logical/virtual disks) to adapter world wide port names.

In the VMware world, zoning is sometimes mis-named (IMHO) as being the same thing as LUN masking. ('zone your LUNs')

Before working with VMware, I have never heard that before and only recognized zoning[/b] as being a feature on the Fibre Channel switches. It is a means to control which ports in a fabric can talk to each other.

These days so-called initiator-based zoning[/b] is fairly standard and even has made it as a recommendation by Compaq/HP's StorageWorks Division. It means that there is usually a one-to-one relationship between the server/host (=SCSI initiator) and a storage array (=SCSI target).

Reply
0 Kudos
Jan_MS
Contributor
Contributor

Thank you for your responses!

Regards,

Jan

Reply
0 Kudos
jasper9890
Enthusiast
Enthusiast

I seem to be having a problem similar to this but am having trouble narrowing down the problem if it's only on a certain datastore, certain guest configuration, etc.. Anyone have any tips on troubleshooting this?

I have the navisphere agent installed on each of the ESX hosts at the suggesting of a EMC installer, could this be causing the problem?

Has anyone had this affect machines while they are being snapshotted, like they would crash or coredump? I think i'm having this on a few solaris 10 vm's.

Reply
0 Kudos
ashlez123
Contributor
Contributor

Hi,

This could be classed as bringing up an old thread so sorry if thats the case. I was having exactly the same problem with VCB about SCSI reservations. It turned out that the cable between our VCB Proxy and Storage switch was dodgy. Seems it was the cable causing my issue.

Reply
0 Kudos