January 18, 201214 yr My server rebuild is in progress. Everything seems to be going very smoothly, including the passthrough of the MV8. Johnm, your guide is absolutely invaluable! The issue I'm running into is with RDM for my unRAID cache drive hooked up to a SATAII port on the mobo. I followed the guide to a T; edited the unRAID VM config, went to add an existing virtual drive, picked the .vmdk file I created, but I just can't get my hard drive to show up within unRAID. Am I missing something? Are there any troubleshooting suggestions? Thanks! Edit: I should mention that this is not a new drive but my cache drive previously used on my non-virtualized unRAID box.
January 19, 201214 yr Author So I'm at a loss. First I couldn't get my cache drive to even show up in unRAID. I solved that problem by changing the (2nd) LSI Logic controller from Parallel to SAS. Then I was able to at least try to assign "sda" but the unRAID GUI keeps going back to unassigned. I unplugged all ESXi-related drives, rebooted directly into unRAID, and everything is as it should be. Can I not do a raw device map when I'm already passing through an entire SAS card? Do I need to get a 2nd SAS just to be able to run a cache? Anybody? Edit: Hmm... for whatever reason I had to use vmkfstools -z instead of -r. That fixed me right up.
January 19, 201214 yr From my research and playing around with ESXi (I'm very new to this so could be wrong), I believe ESXi can only RDM a remote storage device, otherwise you have to do pass-through (two different animals from my understanding). Whereas VM Workstation can RDM a physical drive. So you need to pass through the SATA controller. Usually the motherboard has more than one controller (2-4ports on each) that you can pass through. Finding which is which may be the harder part. if you have ESXi installed on USB you should be ok to play around with pass-through of the motherboard controllers. Keep us posted as I hope to do this myself one day (my current hardware doesn't have VT-d so I can't do pass-through at all).
January 19, 201214 yr If i recall, My instructions have you create a virtual or "Raw disk" RDM. you might have to try a physical or "passthrough RAW disk" RDM. by doing that, you are using the same level of SCSI/SAS passthough communication though the Guest. I don't remember if I mention that as an option in my build notes. this page should have instructions for that: http://www.vm-help.com/esx40i/SATA_RDMs.php the instructions for 4.01 and 5.0 should be the same. This might also fail if your datastore is on a NFS share on a differant box or on a Local SAS controller. the datastore drive (usually) needs to be local (ISCSI Target might also work since it is seen as local). [this is not 100% proven but it has been reported as a possible issue]
January 19, 201214 yr Author Correct, your build notes use the -r option, which for whatever reason just didn't working for me (v 5 maybe?). No clue what the real world differences are in an unRAID environment, but I'm just happy its working!
January 19, 201214 yr Author Yeah sorry, I went the edit route before I got any replies. First, I overlooked the fact I needed to change the SCSI controller LSI Logic from Parallel to SAS (the newly created one when I added the disk). Then I used the "-z" switch instead of the "-r" switch. When using -r, the drive would show up as an option to assign, but it wouldn't stick. Also, it lacked the long model/serial number that is usually passed along with the drive. Once I recreated the .vmdk with -z, everything just worked!
January 20, 201214 yr Yeah sorry, I went the edit route before I got any replies. First, I overlooked the fact I needed to change the SCSI controller LSI Logic from Parallel to SAS (the newly created one when I added the disk). Then I used the "-z" switch instead of the "-r" switch. When using -r, the drive would show up as an option to assign, but it wouldn't stick. Also, it lacked the long model/serial number that is usually passed along with the drive. Once I recreated the .vmdk with -z, everything just worked! I missed the edit. Yes, that i what I meant by the passthough RMD to use the -Z. I recently recall a thread talking abut similar issues and how some motherboards do not pass the serial number and cause issues. I wish I has saved the link. Since this was split out of the atlas thread, I'll post a link back to this thread in the writeup for others. EDIT: I found that thread discussing the -r versus -z http://communities.vmware.com/message/1711875
January 20, 201214 yr Author So strange! We have the same mobo! Thanks for adding to your guide. I might be a fluke, but at least it's an option others can try if they run into a similar issue.
January 20, 201214 yr Yeah.. I think I see the issue. I am going to guess that because you're adding this RDM to unRAID, It needs the -z switch for the SCSI command support. Also, you have to add it as an LSI controller, otherwise it will bomb out. It looks like you might have missed the first time around. Windows tends to require the -r switch. I am sure this is discussed somewhere else on this forum, I just don't recall seeing it. Glad it is working and now documented for others. [*warning* This makes me think that an RDM passed like this might blow up in unRAID 5Beta13 & 14 because of the LSI issues.]
January 20, 201214 yr Author Well we'll probably know soon enough because I am running b14. Does that mean when my m1015 comes next week, I should either hold off on installing it or downgrade unRAID?
January 20, 201214 yr Yes, do not run an LSI based card in beta 13/14. your drives will redball when you spin them up from being spun down.. that's also how to test the RDM passthrough thing. spin down the drive then try to copy data to it. see if it redballs instead of waking up.
January 20, 201214 yr Author Ok I'll give that a shot. My cache never spins down because I still haven't moved Plex off of unRAID and onto a Windows VM, so I'll stop any processes and give it a shot. I've never gotten a redball before. If it happens, do I need to do the trust my array procedure, IIRC?
September 17, 201213 yr Hi I just RDMed my cache drive (256GB Corsair SSD) on my ESXi box, to use it as a Cache Drive for unRAID. i used this method vmkfstools -z /vmfs/devices/disks/vml.01000000003132313438313030303130303134353430324638436f72736169 unRAIDCacheRDM.vmdk -a lsilogic I then attached it to the unRAID VM, and the server booted. when going to the webpage of unRAID (http://tower), and i go into Main, I get three HDDs in my dropdown list. 2 of them are the actual 3TB SATA drives I have installed, and the third one is showing up as VMWare_Virtual_IDE_Hard_Drive_0000000000001 (hda) is that normal? did it work? Update Please ignore the above. It seems that one is only the vmdk created (my USB image). I'm on the unlicensed free version. I just purchased my pro edition of unRAID, waiting for the product key to register.
September 19, 201213 yr Hey.. you need to use the -Z to make a physical disk. then you need to add it to the unraid VM I'll just steal this from another thread.. after you create the -z rdm... follow this. [...] ---- SUCCESS!!! Followed the following in combination to johnm's RDM instructions! Taken from bryanr's thread "UnRAID on VMWare ESXi with Raw Device Mapping" For each of your physical disks: Add... Hard Disk Select the type of disk to use: Use an existing virtual disk Browse... datastore1 -> UnRAID -> diskname Virtual Device Node: scsi(1:0) Mode: Independent / Persistent (it should now appear as a Mapped Raw LUN in VM hardware list) Re-configure the SCSI controller that was added for you Change Type... LSI Logic SAS SCSI Bus Sharing: none as i recall, it will say "Harddisk Mapped LUN" in the VM properties tab once you get this far. the problem is a straight RDM passthrough is not presenting a serial number for unraid to mount it. it was that or the bus type is just invisible to it.. I forgot.. It might also work if you reconfigure the controller to IDE.. I did this to pass a VMDK that was on an NFS/ZFS raid once.
December 2, 201312 yr I am in the same situation as kaiguy but unfortunately everything didn’t just start to work. I am using an Intel DQ77KB motherboard running ESXI 5.5. I too first created the vmdk using the –r and had the exact same issue with it not sticking and reverting back to unassigned. I then tried using –z and have only had PSODs since. Attached is a screenshot of the PSOD if that helps. Would really appreciate a solution as I have spent the entire day trying to find resolution to no avail.
December 2, 201312 yr Hopefully this saves someone time: the Intel DQ77KB coupled with ESXI 5.5 was incompatible. I spent probably 12+ hours trying to no avail to get it working. The integration of the Intel NIC drivers was successful as was the installation of unRAID. However my luck changed once I wanted to use a cache drive that was connected to the onboard controller. PSOD abound. I believe I tried everything. ESXI 5.1 (after integrating the NIC drivers) works perfectly.
Archived
This topic is now archived and is closed to further replies.