Disk Enclosures
1753325 Members
5437 Online
108792 Solutions
New Discussion юеВ

Re: Ownership changes on A/A EVA Controllers

 
SOLVED
Go to solution
Mark...
Honored Contributor

Re: Ownership changes on A/A EVA Controllers

Hi

Patrick - ref part of your answer above -

" If to many IOs are received by the "wrong controller", the LUN ownership is changed. This threshold can be set only by SSSU. "

Please can you tell me what the command is to alter the threshold as I have just been looking through SSSU commands for V9.0 and I do not see which one it would be ???

Thanks,

Mark...
if you have nothing useful to say, say nothing...
Del_3
Trusted Contributor

Re: Ownership changes on A/A EVA Controllers



Peter, Thanks for the good info but there are plenty of ppt's and docs that refer to Master and Slave controllers on EVA XL's. Maybe we should refer to Owning and non-Owning.

Uwe, Again you win the prize. This the answer I was looking for. In particular, I have been pouring through customer EVAPerf data and came to the exact conclusion of your KB reference for ESX 3.5. Now I have the proof and can proceed.

Mark, I have a presentation from 2007 that suggest it was considered to allow admin (SSSU) control of the Implicit LUN transfer but was never implemented.

Thanks again all.
Uwe Zessin
Honored Contributor

Re: Ownership changes on A/A EVA Controllers

The EVA _does_ have a master / slave controller, but it has a different meaning. The master controller, e.g. is responsible for talking to CV-EVA.
.
Del_3
Trusted Contributor

Re: Ownership changes on A/A EVA Controllers

Well that may be but from a 2007 Tech Forum presentation on EVA Performance we have:

"Vdisk access via the Master controller

All read and write requests are sent to the Master controller
Reads are serviced by the Master controller without the slave controllers involvement
The only data that moves across the mirror port between controllers is write data being mirrored to the Slave controller├в s mirror write cache"
Uwe Zessin
Honored Contributor

Re: Ownership changes on A/A EVA Controllers

I understand, sigh. Unfortunately there wasn't a clear use of technical terms from the very beginning (a lot of stuff was completely kept a secret) and even today people use different / wrong terminology.

And I am not even talking about cross-product or cross-vendor terminology.


> control of the Implicit LUN transfer

IMO, another example of flawed terminology (not your fault). A controller can own a virtual disk. A virtual disk is then presented on this controller's hosts ports into a host's SCSI LUN address space (1..255). The non-owning controller also presents two SCSI LUNs on its host ports - that makes 4 different SCSI LUNs on 4 targets.

(Could be worse: some EMC arrays/user interfaces use the term "LUN" in two places - for the analog of a virtual disk and for the SCSI LUN. I think I better stop here...)

Oh, wait... I have a dim memory that in the early days of XCS it was sometimes necessary to disable to automatic ownership transfer of virtual disks. At that time it was possible through a separate interface for service people.
.
Rampy(Venu)
Frequent Advisor

Re: Ownership changes on A/A EVA Controllers

Yes Uwe that was done from field service page by setting the debug flag's.