Situation:
We have 8 esx servers with 2 vmhbas each: 16 vmhbas in total. We have 3 SAN locations for storage (2 FA sets on a symmetrix, 1 clariion, all fibre). Lets call them target C, S1 and S2. For most of the vmhbas (13), the order in which these SAN locations are listed (SCSI Target 0, 1 and 2 respectively) is: C, S2, S1. No problem so far. But there are 2 vmhbas with the order: C, S1, S2 and there is one with order: S2, S1, C. As we have a large number of datastores / LUN's, this is becoming less and less transparent. I would like to have identical SCSI Target orders on every vmhba, so I can easily script a check for misconfigurations (such as a LUN visible on only one hba, or on 7 out of 8 esx hosts).
Question:
How does ESX determine what SCSI Target ID number to assign to each SAN storage location? Can I manipulate this behaviour?
In the mean time, I will hardcode the "SCSI Target number <-> actual target" translation in my script.
Hi Hugo,
We have the same situation here. we think it's because of the order the storage arrays have been attached to the ESX hosts. but we have no problems at all with the attached storage.
The scsi targets you see are referring to the storage arrays.
So what you can try is to remove the zoning and the masking for those 3 ESX servers and recreate the zonig and masking on a array basis in the same order as the 5 other esx servers.
another option for you could be: is to kick of a script wich looks in the ESX hosts log files for problems
or use the "esxcfg-mpath -a" command to compare the esx hosts connected storage
Okay.
We don't have any problems with the storage either, but it would be more clear which path is which storage array if it was consistent.
In the mean time, I have hardcoded the relationship between the esx host + path (vmhbax:y:z) and the storage array in my script.
Hello,
Generally it is first come first serve.... If you have different hardware in each then that could be attributed to it. It is vmhbaBest regards,
Edward L. Haletky
VMware Communities User Moderator
====
Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.
CIO Virtualization Blog: http://www.cio.com/blog/index/topic/168354
As well as the Virtualization Wiki at http://www.astroarch.com/wiki/index.php/Virtualization
Hugo, did you ever find a solution to the mixed up SCSI targets problem? We are having the same problems using different SAN vendors resulting in different SCSI targets all mixed up between ESX hosts in the same cluster and between HBA internallty!
To solve this, you should use persistent binding. That way, the targets will retain the same T number, across boots, regardless of discovery order.
How to do persistent binding on an ESX 3.5 environment? I heard it wasn't available anymore...