iscsi configuration...

[goof1427]

thanks for your time.

I am trying to configure iscsi from the NetApp to ESXi 4.1 host.  I have created the LUNs (as vmware), the initiator group and added the iqn of the iscsi software adapter of the esxi host.  on the esxi host i have added the filer IP in the dynamic discovery tab and after a rescan it has populated the full iqn in the static discovery tab ( i assume this is good communication). I disabled CHAP because when it was enabled it failed according to initiator reports from fileviewer.

I just dont know what I am overlooking or not doing...any help or info will be greatly appreciated.  I have talked to netapp support and we even tried to go directly to the windows server 2008r2 and that was unable to connect as well.

Another question that comes into mind that is should i go directly to the windows box? instead of adding it to esxi host?  doesnt that just put on another layer for failure? if the host goes down there goes the connection?  but if it vmotions over to another host does iscsi connections come with? or does it need to be setup on every host?  And is there ontap dsm for windows since it has better features and benefits then the standard windows iscsi software.

thanks again.

[rorzmcgauze]

When you say you have added the filers IP address in the discovery tab, did you add the ip address of the iscsi interface(s)? or as I think you may have did you add the e0m management IP address?

this needs to be the ip address used for iscsi (eoa, e0b, vif0)as this is for iscsi communication. You would only add the e0m address into the VSC.

[goof1427]

The issue was that I was trying to mount it to the vfiler itself but it should have been added to the one of the vfiler units that was enabled for iscsi.  I was able to this via putty.  What I dont understand is why we can't choose what vfiler unit to create the LUNs on through the onsystem manager.  I guess I will have to learn more CLI.

another question is why not just create SAN volumes instead of LUNs for esxi or windows? aren't they the same thing?

[goof1427]

The issue was that I was trying to mount it to the vfiler itself but it should have been added to the one of the vfiler units that was enabled for iscsi.  I was able to this via putty.  What I dont understand is why we can't choose what vfiler unit to create the LUNs on through the onsystem manager.  I guess I will have to learn more CLI.

another question is why not just create SAN volumes instead of LUNs for esxi or windows? aren't they the same thing?

[goof1427]

The issue was that I was trying to mount it to the vfiler itself but it should have been added to the one of the vfiler units that was enabled for iscsi.  I was able to this via putty.  What I dont understand is why we can't choose what vfiler unit to create the LUNs on through the onsystem manager.  I guess I will have to learn more CLI.

another question is why not just create SAN volumes instead of LUNs for esxi or windows? aren't they the same thing?

[goof1427]

The issue was that I was trying to mount it to the vfiler itself but it should have been added to the one of the vfiler units that was enabled for iscsi.  I was able to this via putty.  What I dont understand is why we can't choose what vfiler unit to create the LUNs on through the onsystem manager.  I guess I will have to learn more CLI.

another question is why not just create SAN volumes instead of LUNs for esxi or windows? aren't they the same thing?

[wes_beckum]

I am not completely positive on this, but you may want to try to create the LUN as a Windows LUN and then present to the VM through VSphere as an RDM.

Wes

[goof1427]

I talked to netapp again and they point back to it being a windows issue saying that the netapp is setup as it should be.  after some more scouring I found this kb article on microsoft website which doesn't help for anything.

http://support.microsoft.com/kb/976072/

[goof1427]

i did a pktt trace and I only see connections being attempted from my esxi host.

ISCSI: New session from initiator iqn.1998-01.com.vmware:xxxxxxxx

here is what i get when trying to connect from windows:

iscsi.warning:warning]: ISCSI: network interface e0M disabled for use; discarding incoming

connection from initiator IP addr "windows IP..."

dont kno why it is trying to connect to the e0M interface.

[enrjfitz1]

Check your cabling, both physiocal and VMware Vswitch.

[goof1427]

the cabling is set.  and i am able to connect to the IP address for NFS datastores.

i did see this article and it somewhat makes sense.  that since all the ports and interfaces are enabled for iscsi it doesnt attach to a portal group that has a network route... so they created an ACL?

http://vikashkumarroy.blogspot.com/2009/03/iscsi-error-for-netapp.html

[goof1427]

another call to netapp but this time with a network engineer on the call as well.  we wanted to show that this isnt a windows or esxi problem...but something on the filer is not set.  I tried to do a telnet from the windows box to e1a IP address on port 3260.  failed.  works fine on port 111.  net eng tried to do the same from a switch...same result. failure.

Trying 12.234.111.11., 3260 ...

% Connection refused by remote host

when I looked at 'iscsi portal show' on the filer ...result was no available portals

so the million dollar question is how come it is saying that when the port is listening and iscsi service is running.  how do i create a portal?

[goof1427]

created a portal group.  what I am failing to recognize is the way the system was setup wasnt complete.  even though isci was enabled for all interfaces for whatever reason the system didnt see them as a portal.  I enabled another NIC and was able to create a portal group and now the filer says it sees 1 portal..the one I just created.  however it is not cabled in.  I will be going to the datacenter tomorrow and cabling it up.

i will admit i dont kno much.  but wat disappoints me is that netapp tech support hasnt been helpful.  that was my case from the begining asking how come no interfaces are showing for iscsi and they said it was normal. 

well its a step closer.  lets see how tomorrow goes.

[enrjfitz1]

Hi, I'm assuming that the Windows Server 2008r2 box is a VM running inside VMware?

Also, what is the iSCSI connection for? When I have configured virtualised MS Exchange etc I have always used the iSCSI initiator directly to the VM and not connected to the hypervisor host. If the VM is setup in this way, any relocations to any other host systems(eg failover) will be carried over.. you just need to ensure network connectivity before it happens so you know you can rely on the process.

Are you using SnapDrive for you iSCSI LUN's? It's not required but is the recommended way of snapshotting LUNs to ensure application consistancy

[goof1427]

jason - thanks for the quick reply.

yes it a windows 2008r2 box running on a VM in esxi 4.1.  the connection is for LUNs that will be file share LUNs for the company.  like a 5TB LUN.  I was thinking the same thing, to go directly from the 2008r2 server to the NetApp however one of the techs from netapp support recommended I let esxi handle everything.  Also isnt there a better iscsi initiator software than the default one in windows.  NetApp has DPM but it is not supported on esxi 4.1. 

we will be using SnapDrive.

[enrjfitz1]

Both options will work:

You could let ESX handle all and just provision a second VMDK file to the Windows VM that is to be used solely for Windows data shares. Format it under Windows and all sorted, no messing around with iSCSI, it all just comes from the VMware datastore (for which you could use NFS). The downside to this though is SMVI would not guarantee application (Windows OS) consistancy if you had to do a SnapRestore of the shares volume.

Using the iSCSI initiator from within the VM, using SnapDrive you can have application consistant snapshots as Windows will dump all RAM to disk before Snapshotting. I've always just used the Windows iSCSI initiator, it connects fine and SnapDrive integrates well with it. To be honest, I wouldn't see why you would need anything else.

[goof1427]

the problem with using NFS on ESX is the limitation of 2TB...

I am going to try the route of having it setup directly on the 2008r2 box.  like you said I want conistency.  And the problem so far from there is that the connection fails when I put in the storage IP or the target (netapp) iqn.  so I gotta figure that out.

[enrjfitz1]

I'd check the following:

1. iSCSI services are running on Windows server (reboot usually a good idea)

2. LUN initiator group includes the iqn of the Windows Server

3. LUN is mapped to correct initiator group

4. The Windows server can ping the IP of the NetApp filer

5. Add (quickconnect) the IP of the NetApp filer - it should connect if the above are correct

6. Use Disk Manager in Windows Server to format the disks

[goof1427]

i went back to my training documents and I have noticed there are no iscsi interfaces listed however all the interfaces are enabled for iscsi if i look through fileviewer.