ISCSI Volume cannot open on both hosts simultaneously!?

[myk3]

Hi all,

I have ESX 3.0.1 installed on 2 hosts, which are both open in the VIC.

I have created the ISCSI adaptor in the network section, I have enabled ISCSI and entered the target IP address on both hosts.

firewall is also enabled for the ISCSI client.

on the ISCSI SAN there is a volume for access of only the 2 hosts which is 400gb in size.

If on host 1 i scan for disks it find the 400gb volume and it is added as storage, great all works fine!

NOW if i go to node 2 and scan for disks it finds the 400gb volume and adds it to storage.

NOW if i go back to node 1 and try to install a VM on this storage it says cannot find storage ISCSI.

If I go back to host 2 and then try to install on the ISCSI storage it will work fine but it says that it is for single access not multi access.

This if i go back to host 1 rescan i can add a VM to host 1 but then the ISCSI storage is then unavailible on host 2 again.

Please can any one advise what the problem is or how I can go about fixing this issue?

Thanks a lot

Mike

[weinstein5]

Have you created a VMFS datastore? It is the datastore to whoch you will store the vortual disks used by the VMs and is filesystem that is sharable between hosts - if you want to access it using an RDM than, yes only one host can access it at time

[AntonVZhbankov]

What iSCSI array do you use? Check if it supports multiple connections per target.

[myk3]

hi there yes i have created the VMFS datastores on both hosts connecting to the same Volume/LUN on the ISCSI SAN, its a QSAN 210C or something like that, sorry im not at work at present so cant tell.

so is the fault that the SAN dont support multi connections per target? or is there something im still doing wrong?

thanks

mike

[kjb007]

What kind of SAN is it? Yes, the SAN will have to support multiple host access for the LUN.

-KjB

[mcowger]

This particular device says it supports "Up to 16 hosts clustered for one volume", so it would be fine.

--Matt

[Dirac · Accepted solution]

1. Check your switch zoning and make sure that both hosts WWNs and the Storage controller WWN are in the same zone.

2. At the array level, check the LUN mapping to ensure that both initiator ids(iqns) are mapped to the same LUN

3. If there is an external metadata server, which I doubt unless you're doing a high bandwith operation such as muti-user video editing, make sure that the mapping/grouping is as expected.

4. Check the LUN masking to make sure that neither initiator is masked from seeing the LUN

-Fred

[myk3]

thank you kindly Sirs!

very silly of me really, i deleted and recreated the LUN to allow all hosts using the (*) attribute.

Qsan is very limited config and i could not find a way to specify a LUN to 2 host iqn addresses without using the all symbol.

thanks for all, as its only a test server its fine,

thanks again

mike

[kjb007]

Don't forget to mark the question as answered and give points for helpful/correct posts.

-KjB