Attaching more than 1 san target

[jetlee]

Hi All

My first post, be gentle ..

Ive successfully installed 3 diskless clients using Win 7 and AoEDsk (which is quite a bit better than the original AoE Driver from a performance perspective). The main disks are hosted on LVM Raid 5 (mdadm) volumesand the PXE environment is tftpd-hpa and isc-dhcp-server on a Debian Wheezy server, which is chainloading iPXE via PXE.

At this point Im presenting a single LVM volume as a single drive to the initiator, which works flawlessly. I have recently got my hands on a rather small (64GB) ssd, which Id like to present as a purely swap volume, to maximize the performance if a machine is swapping. Currently the swap files exist as a regular swap file on the main volume and I have no real performance issues, but you know how it is, if you can make it snappier, you basically do try ..

So this leaves me in a predicament.. I cannot "late bind" the swap to a voluem I can present in the startup scripts in Win 7.. so the most logical solution is to present the swap as a second drive to the OS at iPXE boot level, so that the OS is aware of its existence .. but for some reason this is not working ...

Ive noted a similar (although iScsi) post here http://forum.ipxe.org/showthread.php?tid=7518, but it has had no response .. from what I gather in the iPXE Documentation on the sanhook command, there should be support for multiple targets, but only 1 is allowed to boot from.

From what Im picking up, it appears as though in order to recognize the drive, it requires an iBFT entry, and apparently this functionality only exists for iSCSI multipathing, but any other drives are avialable by int13 (??) ..

Does anyone have any idea, or input around either another workaround for my issues, or a way to attach multiple targets early enough in the boot sequence to allow successfull boots ??

Cheers

Jet

[robinsmidsrod]

As mentioned in the response to the other thread you referenced, you can have only one volume described at boot. I'm not sure why, but you could try to disable the removal of other volumes and see how it works. Check the int13_describe() function in the source for the line to disable and recompile. Please do tell how it works out for you.