Stradle, looks like that backplane isn't supported by ESXi.  When you're in vSphere, under Configuration -> Storage -> Devices, do you see your hard drives?  I'm going to assume no, based of of your fdisk -l report.

 

Does your motherboard support VT-d?  Can you pass through the backplanes direct to the unRaid VM?

 

I'm actually in the process of writing up my guide for getting unRaid to work in ESXi, which varies slightly from this guide as mine will focus more on passed through hardware.  I should have it published in a day or two, maybe it will help.

 

Hi, no the HDD's are not showing up, but neither is a seperate hard drive that I connected completly outwith the backplane which is strange.

Would this be the cause of it not allowing me to use the 'mkdir' command?

 

I will need to check the VT-d support, not sure about that, but the hard drives are functioning currently in my Unraid setup just not when I use ESXi.

I look forward to trying your guide out, but any other ideas until then I would appreciate, cant believe how early in the current guide that I am stumped.

 

 

I've experimented with running unRAID in a ESXi VM and like it, but didn't like the idea of having all my storage on my ESXi system as well.  What I would really like is a separate machine that manages all of my storage with chunks of storage being allocated to different VMs as required.

 

Yesterday I decided to run an experiment by setting up an Ubuntu server as an iSCSI target, using the ESXi system containing the unRAID VM as the iSCSI initiator, configuring ESXi to map the drive on the Ubuntu server to the unRAID VM, and then adding the drive to my unRAID array just as if it was a local drive.  I'm happy to say that it worked just fine!  I rebuilt parity, checked parity, copied a movie over to the remote drive and played it back on another machine without any problem.

 

Beacause this was just a quick proof of concept, I set up the Ubuntu server acting as the iSCSI target as a VM on another system instead of on its own hardware, but even with all of the virtualization and network overhead, everything seemed to work fine.  The parity build ran at a speed of 59169K/sec and the parity check at a speed of 85635K/sec which seems respectable to me.  This was with the parity drive and 1 data drive on the unRAID VM machine, and 1 data drive as an iSCSI target on the Unbuntu server machine.

 

I suppose the next thing I should try is to configure all 3 drives as iSCSI targets on the Ubuntu server and see what happens.  At some point the network bandwidth will become a bottleneck, but that can be migitated by setting up a separate subnet for the storage related traffic.

Has SK or anyone tested the LSI 2008 Chipset cards? Do they experince the same issue as the 1068E chips with regards to no temp readings/spindowns

unRAID 5.0b6a Port 80 "Main" tab, executed a "spin down". Cache drive on the SAS 2008 Controller spun down, but non of the 7 DATA drives??? To add from the unMENU "myMain" page it showing all 14 disks are spun down?

 

What is the spin down command I can try from the console?

This is a great thread, I think I've read through it at least three times in the last couple of days.  I want to see that I understand properly, can someone verify this for me.

 

To install UnRAID on an ESXi VM with hard drive spin down support there are two options currently.  Both requires a motherboard with Intel VT-d support or AMD IOMMU

• Use AOC-SAT2-MV8 on a PCI-X bus and pass the entire PCI buss to UnRAID
  
• Use a SATA card that will run with the mptsas as posted by SK.  This requires patching every version of UnRAID for the modified drivers and plugins will also require modifications.
  

 

Is that right?  So if I want to have hard drive spin down, don't want to lock myself into a community created mod, and would like expansion beyond the motherboard it's AOC-SAT2-MV8 and pass the whole PCI buss through to UnRAID.  If so that makes it impossible to have a SATA expansion card AND a PCI Tv-tuner (for a different VM).

 

If the idea is to support UnRAID Array, a drive for the VM's, and another drive for Recorded TV / Temp storage.  All you really are guaranteeing given a typical 6-SATA motherboard would be 4-drives for UnRAID.  I guess with a Pro licence and a Cache drive you could possibly record tv to the UnRAID array... but that doesn't buy you anything since you just move the TV-drive to Cache and don't gain any extra array storage or protection on the cache drive.

 

Someone help me out here.  I'd really like to have a combined ESXi box for Sage/UnRAID/ and still have some room for a VM or two for hosting other services.

Hopefully I've just missed an easy solution for *supported* SATA extenders and the ability to still use 1 or 2 of my PCI slots.

You're missing an option or two. They're both quite easy to achieve.

 

1) Disable unRAID controlled spin-down and switch over to drive-controlled spindown by using hdparm -S.

2) Disable unRAID controlled spin-down and use your own idle-detection script then issuing hdparm -Y or hdparm -y.

3) Disable unRAID controlled spin-down and use sg3_utils/spindown to handle it [http://code.google.com/p/spindown/].

You're missing an option or two. They're both quite easy to achieve.

 

1) Disable unRAID controlled spin-down and switch over to drive-controlled spindown by using hdparm -S.

2) Disable unRAID controlled spin-down and use your own idle-detection script then issuing hdparm -Y or hdparm -y.

3) Disable unRAID controlled spin-down and use sg3_utils/spindown to handle it [http://code.google.com/p/spindown/].

 

That's wonderful news.  So since all of those don't require direct disk access that makes it easier to use.  I don't know much about those methods but I'm guessing at least one of the three is easy enough to get setup on unraid ... maybe even a plugin if this catches on  

 

So if I'm gonna use one of the above UnRAID unaware methods what are the restrictions on the hardware?  

For example if using an LSI card is the 'SK-patch' still necessary?  So if I want to run unpatched UnRAID with unmenu plugins (As I am now) what hardware options does this open up for me?  Basically anything that's supported by vSphere?  Such as the B10i devices that I saw posted earlier?

 

Also those methods don't require passing a whole buss through to the UnRAID VM correct?  So I can have a PCI SATA adapter and still have a TV-Tuner in another PCI slot.

Does this also mean that UnRAID is working with virtual drives?

 

Sorry for all the questions.

I tried to compile some information about this thread, but everything is a bit messy with contradictory statements. Can please someone answer some questions?

 

1) So, using a supported HBA with ESXi (e.g. LSI chipset cards), we can pass disks directly to an unRAID VM, correct?

 

2) Which mode is preferred, LSI HBA emulation or paravirtualized with VMWare tools modules?

 

3) Is device pass-through using VMDirectPath hassle free? Can we map a AOC-SASLP-MV8 card to a unRAID VM successfully?

 

4) Can any of these methods be used to spin down the hdds?

 

Thanks a lot!

I tried to compile some information about this thread, but everything is a bit messy with contradictory statements. Can please someone answer some questions?

 

1) So, using a supported HBA with ESXi (e.g. LSI chipset cards), we can pass disks directly to an unRAID VM, correct?

 

2) Which mode is preferred, LSI HBA emulation or paravirtualized with VMWare tools modules?

 

3) Is device pass-through using VMDirectPath hassle free? Can we map a AOC-SASLP-MV8 card to a unRAID VM successfully?

 

4) Can any of these methods be used to spin down the hdds?

 

Thanks a lot!

 

I can answer a couple of these questions from my own setup.

 

1) Yes, with a supported controller (one the ESXi sees the drives) you can map the raw drive to the VM.

 

2) I paravirtulaized the SCSI controller in my VM. It has been working fine for me without any issues.

 

As for the other two questions, I have not worked with the specified controller so I have not experince with it.

 

I can tell you that my UnRaid server console indicates that the drives "spin down" after the allotted time of no activity, but I have taken the time to verify if the physical drives actually spin down or if it is just the software indicating they are idle.

 

Rod

Thank you very much for your answers, Worlok. Can you post your config just so can I have some perspective?

 

ChaOConnor have you had any luck with a write up on this?  I would be very interested to see something written that didn't have all the "uncertainties" that I have been reading in this thread.  TIA.

My build is completed, stable and successful (excluding spin down which I can probably do manually but don't really care enough about it... 120W's while other VM's are idle but it's a beast of a build so I'm happy with it). I'm going to type up a UCD thread soon on my experiences etc. just a matter of me taking some pictures and typing it all out. My build got pretty vast and serious as time was going on not to mention 2 months of fighting with eBay . It was quite the journey.

 

Needless to say, I chose to pass my controllers through instead of individual drives (2x BR10i). So far everything is going swimingly well. I'll post a link off to the other thread from here once I get a chance to type/document it all. I had a friend who was interested in this so I'd have to type it up for him anyway .

My build is completed, stable and successful (excluding spin down which I can probably do manually but don't really care enough about it... 120W's while other VM's are idle but it's a beast of a build so I'm happy with it). I'm going to type up a UCD thread soon on my experiences etc. just a matter of me taking some pictures and typing it all out. My build got pretty vast and serious as time was going on not to mention 2 months of fighting with eBay . It was quite the journey.

 

Needless to say, I chose to pass my controllers through instead of individual drives (2x BR10i). So far everything is going swimingly well. I'll post a link off to the other thread from here once I get a chance to type/document it all. I had a friend who was interested in this so I'd have to type it up for him anyway .

 

+1

will really like to see this write-up.

I have VMWare also and using pass thru with BR10i cards to which after many weeks of wait on the BR10i cards I got refunded (from ebay) and ordered from another source.

I do have spindown working and placed in the go config the spindown parameters and it's working great.

 

>>excerpt of my go script<<

 

installpkg /boot/webGui/webGui-1.0-unraid-speeding_ant.tgz

installpkg /boot/webGui/simpleFeatures-0.1-unraid-speeding_ant.tgz

installpkg /boot/vmtools.tgz

 

sleep 60

 

#hdparm -S sleep parameter values

#-S1     5 seconds

#-S2    10 seconds

#-S3    15 seconds

#-S6    30 seconds

#-S12    1 minute

#-S24    2 minutes

#-S36    3 minutes

#-S60    5 minutes

#-S120  10 minutes

#-S180  15 minutes

#-S240  20 minutes

#-S241  30 minutes

#-S242   1 hour

#-S244   2 hours

#-S246   3 hours

#-S250   5 hours

#-S0     standby is disabled, will not spin down

 

/usr/sbin/hdparm -S242   /dev/sdq

/usr/sbin/hdparm -S242   /dev/sdp

/usr/sbin/hdparm -S242   /dev/sdk

/usr/sbin/hdparm -S242   /dev/sdj

 

<<end go >>

 

 

I tried to compile some information about this thread, but everything is a bit messy with contradictory statements. Can please someone answer some questions?

 

1) So, using a supported HBA with ESXi (e.g. LSI chipset cards), we can pass disks directly to an unRAID VM, correct?

 

2) Which mode is preferred, LSI HBA emulation or paravirtualized with VMWare tools modules?

 

3) Is device pass-through using VMDirectPath hassle free? Can we map a AOC-SASLP-MV8 card to a unRAID VM successfully?

 

4) Can any of these methods be used to spin down the hdds?

 

Thanks a lot!

 

1) Yes

2) I prefer VMDirectPath for pass the Controller Card though.

3) It is simple, when you have the right mix of hardware.  With the motherboard I have (Intel DQ45CB) I found a few cards did not pass.  I currently pass 3 JMB362's

4) Spindown and temp reading work perfectly on controllers passed via VMDirectPath

Zeron, thanks a lot for the answers!

 

Today I did a "weird science" test! I made a ESXi USB boot device, put it into my server and boot it up. Configured a unRAID VM with usb using Plop Boot CD, then started to make a lot of experiences using my motherboard SATA ports.

 

In the end, paravirtualized SCSI with PVSCSI driver shown better capabilities than LSI SAS emulation; using it, I could even set spindown timeouts to my drives using hdparm -Sx. The performance is also almost identical to the "real world" numbers. If I had a LSI 1068 or a LSI 2008 card here, I could keep it running virtualized without any hassle.

I tried to compile some information about this thread, but everything is a bit messy with contradictory statements. Can please someone answer some questions?

 

2) I prefer VMDirectPath for pass the Controller Card though.

4) Spindown and temp reading work perfectly on controllers passed via VMDirectPath

 

This is what puzzles me - I've read that the br10i card works with VMDirectPath, but if 4) is the case why are there reports that spindown and temp readings don't work with the br10i?....

 

Confused.

I tried to compile some information about this thread, but everything is a bit messy with contradictory statements. Can please someone answer some questions?

 

2) I prefer VMDirectPath for pass the Controller Card though.

4) Spindown and temp reading work perfectly on controllers passed via VMDirectPath

 

This is what puzzles me - I've read that the br10i card works with VMDirectPath, but if 4) is the case why are there reports that spindown and temp readings don't work with the br10i?....

 

Confused.

 

Temps work on the latest beta but spindown does not from all/most reports.  It is the way/differences in how SATA controller cards operate.  Some are different than others, some require a certain command (3MWare I think, among others), and so on.

The spindown commands that EasyMe posted above are compatible with the BR10i. I tested on my server last night after I saw the work done for me ( thanks lol). I was able to tell that they were actually spinning down by powerusage. My server would drop 15-20W's as the drives presumably spun down... then accessing would cause them to spin up again. Wonderful news for me.

 

 

i am about 1/2 way through re-reading this thread, as I finally have all my ducks in a row to get going with this thing. my drives have arrived. need to get the WDIdle run and then test the drives out too. One of them looked a little like it was dropped (the anti-static back was indented in spots).

 

Newegg's got a sale on Adaptec Raid/Jbod cards - might be something to consider?

Good news, Tom's latest build (v5 beta6-d) natively supports spin down for controllers like mine (as I mentioned i have 2 BR10i's). I and others have already gotten a chance to test it and it seems to be working well.

 

One thing I forgot about is that updates to your flash do not carry over to your unRaid vdisk image. Am I doing something wrong or do I need to create my vmdk every time I upgrade something like this? And if no, is there an easier way to upgrading than adjusting you usb drive with appropriate changes, recreating the vmdk files and then reuploading into the datastore?

One thing I forgot about is that updates to your flash do not carry over to your unRaid vdisk image. Am I doing something wrong or do I need to create my vmdk every time I upgrade something like this? And if no, is there an easier way to upgrading than adjusting you usb drive with appropriate changes, recreating the vmdk files and then reuploading into the datastore?

 

hmm interesting didn't think of this - but sounds right. bryanr mentioned something like this in the first post:

"Comments/improvements welcome - particularly around the faff of creating a virtual boot disk image, which seems like overkill?"

 

for now, i'm going to run pre-clears on my 3 drives. it's not ideal to have to plug in the flash and make a vdisk each time you upgrade but can certainly live with it as a "downside" of this configuration.

 

 

I did it through one of my VM's what's the point in having a nice ESXi all-in-one machine if you can't do it all remotely haha

An easy solution to booting directly to USB in ESXi is to create a small virtual disk ask your boot drive, and install Plop Boot Manager (http://www.plop.at/en/bootmanager.html). This will then act as your boot manager, and you can configure it to boot directly to the USB stick. That way you can just upgrade the unRAID flash drive the same as you would on a physical installation.

Thanks fade, I'm going to have to experiment with that since I don't want to have to create a new virtual disk every time I upgrade something lol.

An easy solution to booting directly to USB in ESXi is to create a small virtual disk ask your boot drive, and install Plop Boot Manager (http://www.plop.at/en/bootmanager.html). This will then act as your boot manager, and you can configure it to boot directly to the USB stick. That way you can just upgrade the unRAID flash drive the same as you would on a physical installation.

 

thanks - this will also work!

 

i saw the pre-clears could take 30hrs on a 2TB drive. i've got three of them - time to break out the popcorn