Dynamic Lun Addressing

Try this --

symmask -sid XXXX -wwn XXXXXXXXXXXXXXXX -dir 3a -p 0 add dev 02FA -dynamic_lun

symmask -sid XXXX -wwn XXXXXXXXXXXXXXXX -dir 14a -p 0 add dev 02FA -dynamic_lun

Hi Rahman,

Please provide mapping command also.

yes you do, what do you mean there are not free LUN id ? When you run symcfg list -sid XXX -dir 3a -p 0 -address -list   ..no available addresses ?

map dev XX to dir XX lun=XXX;

Hi Dynamox,

In case of dynamic LUN addressing do we need to mention LunID..?

map dev XX to dir XX lun=XXX;

There are no available free LUN IDs.

Yes exactly.

So we are mapping LUNS using ECC. Could you please provide entire commands to provision LUNS using dynamic LUN addressing in CLI.

Thanks

there are 2 ports per CPU ( ie 4a:0 and 4a:1), each CPU supports 4096 addresses. If you map a meta device that consists of 5 members, each meta member consumes an address. So while your meta looks like one big device, once you map to to an FA it actually consumed 5 addresses. Dynamic LUN addressing helps with operating systems that can't see any devices past FA address 0FF (256). By using Dynamic LUN addressing , you can map device to address 232(decimal 562) and then when you present it to your host, you can specify -dynamic_lun during masking so that device will show up with the next consecutive address for that particular host.

ThanQ Dynamox.

But we are mapping LUNS to those FAs by using ECC. How it possible. do ypu have any idea.?

dynamic lun addressing will not help you if you do not have available LUN id on the FA, you will need to find another pair of FAs, zone your host to them and then provision.

It seems like you need to map the device to another FA port as you have hit the address limit of 4096/CPU limit.

regards,

Saurabh

I have the issue similar.

run symcfg -sid -FA 7f list -address -available, I got the out put

Mapped devices: 208

including metamembers: 2571

Available addresses: 19(s)

for 7F:1

Mapped devices: 196

Including metamembers: 2011

Available addresses: 19(s)

2571+2011 = 4582

If the limit is 4096, how could? and why 19?

Can anyone help

did u figure this out by now? interesting thin though

Are some of the devices dual mapped to both 7F0 and 7F1 perhaps those are only counted once? 

Yes, some devices are mapped to both 7F0 and 7F1. Some of them with same LUN ids, others with different. Are they use double addresses?

to simplyfy, should we only use 0 port not 1 port? Is there any performance issue. It is on VMAX

Another question, should we separate the HBAs by farbic so they do not map to both 7F0 and 10F0 since they only zoned to 7F0 for half of the HBAs. this is in cluster environment for ESX hosts. thanks

well, two ports (0 and 1) share the same CPU, i try to avoid creating port groups that consist of ports from the same CPU.

This paper has some examples how to zone system in single and multi engine VMAX

http://www.emc.com/collateral/software/white-papers/h6340-powerpath-ve-for-vmware-vsphere-wp.pdf