When adding an iSCSI Storage Repository  (SR) to Citrix Xenserver, remember the following checkpoints:

• Use a dedicated interface for storage traffic.
  This is done by right-clicking on a host in the XenCenter GUI, and selecting “Mangement Interfaces…”.
  Create a new interface, giving it a good name such as “iSCSI LAN 10.x.x.x”.
  Set the IP address of the new interface.
  Do this for all XenServer hosts that require access to the iSCSI shared storage.
  Remember that all hosts should have access to the same iSCSI LAN(s).
• Give each XenServer host a meaningfull iSCSI IQN (name).
  This is done on the General tab on each XenServer host.
  Click Properties, and  then change the iSCSI IQN.
  This will help you in identifying each host when allowing their access on the SAN.
• Map volumes to all XenServer hosts.
  On your preferred iSCSI SAN, create a volume and map it to all XenServer Hosts.
  Remember to map to the same LUN # for each XenServer Host. So if using LUN # 10 for Volume_1, use this for all of your hosts, as identified by their iSCSI IQN’s.
  Remember to map all XenServer hosts to every volume you intend to use for “agile” VM’s at pool level.
• Create the new SR on a pool level.
  Right click your pool in the XenCenter GIU, and select  New Storage Repository…
  Select iSCSI. Enter a good descriptive name. Enter the IP address of the SAN controller that has ownership of the vdisk in question.
  Click Discover IQNs. Make sure that the correct controller (which has ownership of the vdisk) is selected.
  Click Discover LUNs, and select your desired LUN.
  Click Finish.

Once you start populating your SR with Virtual Disks, make sure you give them all a name, and a description. This is very helpfull if you have to re-build your XenServer farm.

While adding multiple Network Interface Cards (NICs) to a XenServer host, I needed to re-arrange them so their logical naming (eth0, eth1, eth2 etc…) matched that of their physical order.

Here’s a quick guide how to do that.

1. Before adding your NICs to your XenServer host, note the MAC addresses of each port. You might also want to put this on a label, and affix it to the bracket itself, so you can determine the MAC of the NIC by looking at the rear of the server.
2. Add your NICs to your XenServer host. Note where you position each card, so you know which MAC goes where.
3. Boot XenServer.
4. From the shell (Hit Alt+F3 after boot has finished), type in:
   xe pif-scan host-uuid=<tab-complete your hosts UUID>
5. This will scan for new NICs, and enumerate them as Physical Interfaces (PIFs).
6. Determine the order and UUIDs of the new PIFs, by running:
   xe pif-list
7. This will return the uuid, device (which is actually device-name) and other info about each PIF. Check whether the device (-name) fits what you had planned. Eg. if it says eth2, and that’s what you wanted, just leave it. If you wanted it to be eth1, continue.
8. Remove the PIF from XenServer, by running:
   xe pif-forget uuid=<UUID of PIF you want to remove>
9. Add the PIF again, by running:
   xe pif-introduce host-uuid=<UUID of your host> mac=<the MAC you noted earlier, seperate octets with :> device=eth<x>
10. For example:
    xe pif-introduce host-uuid=8c5bcb04-184b-4523-b870-5d3a5804ca67 mac=00:19:5b:5a:2d:a6 device=eth3
11. After this, your new PIF might show up with unknown device and vendor. While merely cosmetic, a reboot of your XenServer host should fix this.

I was able to use D-Link DGE-528T cards, using the above, even though they aren’t officially supported on XenServer. Well… 2 out of 3 cards worked when placed in a single XenServer host. All 3 cards were detected, and named using the above method. 1 of the cards was unable to recieve an IP for the management interface, using DHCP.

This was tested on XenServer 5.0 update 3.

Thanks to tiris blog for pointing me in the right direction on this.