Personally, I want two parity disks. Although ESXi support would be cool, it is not my top priority, dual parity is.

 

ditto, as much as I would like this to be easier I would much rather see dual parity implemented first.

 

While dual parity is good to have in _future_ as well as some other features available in enterprise/commercial and even SMB storage market what should be possible with ESXi route _now_ is to partition available disks and provide subsets to different unraid VMs.

Say instead of 14+ plus parity disks config to have unRAID VM1 with 7+ disks plus parity and another unRAID VM2 with 7+disks plus parity.

 

From data protection perspective this would allow to reduce raid group size to improve failed disk recovery time and reduce possibility dual disks failures. Obvious price to pay - additional parity disk(s) and more complex config. Also current licensing is per unraid instance.

 

 

+1 for more support on ESXi.

 

I'm currently running UnRAID 4.7 on my ESXi 4.1 Update 1 server, but its far from perfect.  As discussed by others, this means that UnRAID is used purely for storage, and I can have run any other stuff I want (webserver, etc) from separate VMs.

 

Cheers,

Bryan

+1 to ESXi Support for at least the HD Controller Driver piece.

+1 to ESXi Support, currently running unRaid 5.04 under ESXi 4.1, only problems so far are the usual missing temps and no spindown on the drives connected to the IBM BR10i card running using VMDirectPath.

+1 Ditto on the ESXi support, really the only feature i'm interested in right now

+1. I'm also for ESXi VM support, seems to be easier/better scalable then running the VM in unRAID.

 

+1, money in the bank and willing to spend it if this could be done 

 

Having a hard time finding hardware to put a 24 drive system together. All sorts of problems with SAS/SATA controllers out there. We can't keep buying old gear, or stuff that is not being sold anymore. Need Lime to jump in and help support MORE controllers, etc... out there.

Supporting more controllers gets more people excited about unRAID...bringing in more new users. 

Is there any news on this topic?

Hi,

 

I'm currently running 5b12 on esxi 5, pcie passing through my 2 x BR10i controllers. All seems fine so far, spin down and temp work fine this way as one would expect.

 

Would prefer not to have to pass controllers through to allow more flexible controller options going forward, but would need Unraid and esxi to play together in some way to enable spindown/temp (for instance what happens if br10i blows up and i cant get an easy replacement?)

 

Having various vm kernal modules 'out of the box' with the above spin/temp support and optional tools as a package would be ideal.

passing through an entire controller is the most efficient way to manage large quantity of drives and also gives you the best data throughput.

 

to answer your question about dead controller:

Because you ARE passing the entire controller through, if you BRI10 dies.. you shut down ESXi, replace the dead controller with any esxi/unraid compatible controller. boot up ESXi, unassign dead controller, assign new controller.. start unraid... unRAID V5.X will just start. unRAID V4.x you might have to assign the dives again (that's a 4.x thing, not esxi)

 

 

 

much easier then swapping out controller in sxi then reassigning 8 individual drives to esxi and then reassigning to unraid if you kill a controller..also, everytime a drive dies, you have to replace it in esxi then unraid with single drive passthough.

 

with card passthrough, you can just add and remove drives at will and ESXi wont care (or even see them)

not to mention, esxi has a 2TB drive limit. with 3 and 4tb drives shipping, why shoot yourself in the foot storage wise.

What is currently the best PCI-e, PCI-X and on-board controllers to use with unRAID under ESXi? Does it matter? Do they all work the same as running on bare metal when you pass them through to unRAID?

 

Don't you need to have a local physical disk that is not being passed through for ESXi's datastore (holding the actual unRAID install, and other VMs)? So you can't pass ALL of your controllers through to unRAID?

 

What is currently the best PCI-e, PCI-X and on-board controllers to use with unRAID under ESXi? Does it matter? Do they all work the same as running on bare metal when you pass them through to unRAID?

 

Don't you need to have a local physical disk that is not being passed through for ESXi's datastore (holding the actual unRAID install, and other VMs)? So you can't pass ALL of your controllers through to unRAID?

 

 

Well, you could technically boot ESXi off of a thumb drive and then ONLY have unRAID running on your ESXi server which also boots from USB.  In this case you wouldn't technically need a datastore for ESX.  If you want to run anything else though, then yes you need a second controller to present disks to ESX.

 

Look for server class SAS stuff of which LSI cards are almost always a safe bet.  While many different cards typically will work only server class SAS controllers are on ESXi's HCL and are guarenteed to work well.  The IBM M1015 is usually right around $100

 

I have a BR10i in HBA mode hosting disks for ESXi datastores and am using LSI SAS2008 based cards to passthrough to unRAID.

 

Well, you could technically boot ESXi off of a thumb drive and then ONLY have unRAID running on your ESXi server which also boots from USB.  In this case you wouldn't technically need a datastore for ESX.  If you want to run anything else though, then yes you need a second controller to present disks to ESX.

 

You need a Guest setup in order to boot unRAID from USB, therefore you need a datastore.

Also if you would not be running anything but unRAID, why would you want to boot ESX and then unRAID?

you're right if you do it the way that has been mapped out elsewhere on this forum.  In that scenario you have a drive image which is an image of your USB stick, and will be on the datastor.  BUT, you COULD map the usb stick in raw and boot from it, no datastor required.

 

But as you noted, it would be pointless.  The whole idea of doing it over ESXi is so that you can run other VM's which won't interfere with unRAID in any way.

But you could run multiple unRAID's under ESXi using USB passthrough for each respective USB boot-key.  So it's not entirely pointless as you could have multiple unRAID servers running with different drive arrays that operate under different conditions. You would be using an extra drive for parity, but that might provide better drive fault tolerance if desired. 

you're right if you do it the way that has been mapped out elsewhere on this forum.  In that scenario you have a drive image which is an image of your USB stick, and will be on the datastor.  BUT, you COULD map the usb stick in raw and boot from it, no datastor required.

 

But as you noted, it would be pointless.  The whole idea of doing it over ESXi is so that you can run other VM's which won't interfere with unRAID in any way.

I was not referring to a image of unRaid on a vmdk. How/what are you doing in esx to tell it to start and boot the unRAID usb stick? if you dont have a guest? and to create a guest its going to force you to select a datastore. Once you have a guest you could passthrought the usb stick and controllers. Or am I missing something your saying?

I'm not doing it at all, was just commenting on the approach Brit mentioned.  However, his latest post about running multiple unraids could have some merit.  When I have some time (?) I'll try to create a VM booting off the USB stick.

Yes, you can run more than one unraid per esx box. I have 3 on mine. 1 production and 2 test. The test only have virtual drives. While the production has 3 hba cards with real drives.

 

I passed USB drives through for each unraid. And I had to create a datastore for the esxi configuration storage.

 

But yes. You could in theory put 2-4 unraid guests on a single esxi server using DAS storage arrays.

Yes, I run a second one as dev on one and second QA on another, but my point is just one thing, you have to have a datastore.

 

I would love to hear who is running 1 or more VM's without a datastore.

Yes, I run a second one as dev on one and second QA on another, but my point is just one thing, you have to have a datastore.

 

I would love to hear who is running 1 or more VM's without a datastore.

 

Forgot about that part, you are right.  You need a datastore of some sort to store the VM config files on.

+1 on my side too for better unraid support in virtual environment !

so i have a pratical example below. Can anyone here help to tell more about issues i encounter regarding real (or unreal?) issues i have with my unraid vm config on ESXi and how to solve them (change hardware or config?)

 

Here is my HW conf:

asus P8BWS (C206 chipset) mobo with xeon e3-1245 proc

adaptec 51245 raid card

a bunch of 1To and 2 To sata harddrives in backplanes

esxi datastore stored on drives connected in AHCI mode to mobo sata ports.

...

My SW conf is following: ESXi 5.0 on usb drive (with plop iso image for startup) and Unraid USB key in trial mode

harddrives are direct-attached through VT-d passthrough

My unraid is version 5 beta14 - no add-on modules (apart from powerdown script)

 

here are the issues:

- no drive temp: is there a way to have it "captured" from a SMART command ?

- from time to time, either parity or one of the 2 other drives appear in red. I can change the drive and re-use it, so it doesn't seem to be a "real" problem with the drive

- disks make some noise from time to time (like it were broken but i doubt that most of them are)

- don't know if i can benefit from disk spin down ?

 

Can you tell me if these issues are coming from with unraid V5beta14 trial version or if they come from a configuration issue on my side or from a bad hardware choice (adaptec card - C206 chipset) and furthermore how to solve them?

You'll find enclosed printscreens of what i have with my system log.

 

Note that i can use Unraid for my data storage without any pb for a few weeks, but i need to be sure that what i'm seeing in Unraid configuration interface is ok or not?

Thanks in advance for your replies

Pierre

unraid-issues.zip

pegounet, drop down to 5.0Beta12a and re-test, your clearly red-balling on b14. Read the Beta 14 post to understand why, or just drop down to beta12a and re-test.

 

hi madburg

thanks for your advise. I did go back to 12a and still same symptoms occur: (see enclosed printscreen in unmenu and my syslog file)

 

any ideas?

pegounet-syslog-prtscreen.zip

The answer is the Samsung drive.  I upgraded my server to two M1015's and 16 Samsung 2TB drives.  Initially, (running Beta9) when the drives were always awake as I was moving data I had no problems, but once I set the drives to spin down after a short time, most times I went to access a drive, it red-balled.  Going through a divorce, down sizing and then moving, so I just turned it off in hopes that the kernel issues with LSI controllers highlighted in the Beta14 thread would eventually solve the issue.  I had no issues with the previous Seagate drives.  I REALLY hope this issue gets solved or I'm in deep shit!