Hmm.. I‘m sorry I have only „tested“ this behaviour with the power outage I had. I did not stop the array with the GUI or anything. I thought the array is offline during a parity check which is definitely not right.

I forgot I had my array in „auto start“.

 

Nevertheless, it would be really a good idea to be able to run VMs from unassigned devices or cache drives with the array offline.

 

Sorry again = (

Allowing access to array members (cache included) while the array is stopped is probably not going to happen, for various reasons Tom has stated before.

 

However...

 

I see no real reason that we couldn't allow VM's that aren't on the array to operate. It's not like you couldn't set up a VM configuration where all the required elements were on a non array device, but it wouldn't be able to easily share those elements with the VM's currently configured. You would need a separate XML storage location, and an iso location for install media.

 

Maybe a GUI section for array limited VM's, and another for VM's on unassigned devices?

 

As soon as limetech actually takes control of UD instead of relying on dlandon to maintain it, I see no reason why we can't have 2 VM types. However, as long as limetech doesn't actively manage unassigned devices, it's probably not going to happen unless someone writes a plugin.

Unraid noob here, but it looks like this feature is something that I need as well.  Specifically, I have a Zerotier docker running that only uses the cache drive.  If I make a change on the server that requires stopping and starting the array then the docker shuts down and I lose access to the server.  I only have remote access to the server so this is a huge issue.

Unraid 6.8 supports WireGuard, which allows remote VPN access independent of the array status.

+1

I'd also like to see the ability to start specific docker containers. If the data lives entirely on cache would be the condition to allow this. I.E. anything mapped to /mnt/user* would immediately disable this capability. 

 

I think RegEx could probably handle checking VM XML and Docker config for any reference to a non cache directory.

+1

 

This would be very useful!!!  VMs as well as docker containers!

+1

 

this would be a great feature. 

This world make using PFsense VMs a lot less chaotic when bringing down the array. 

I would like this also.

My VM images are not on the cache disk, but on high speed unassigned devices drives, one step further from the array.

 

+1 count me in

+1 from me too! 

Mainly for pfSense, but also because ALL of my VM's are on a separate unassigned drive. I think the majority of users have their setup mostly this way due to the inexpensive SSD's cost and the large performance gains.

+1

Sent from my NX569J using Tapatalk

+1 from me as well. It's really the only thing giving me pause for pulling the trigger on actually purchasing once the trial is up - all my vms/containers/shares are on zfs anyway, so all having an array does for me in the first place is allow me to actually use all the things that have nothing to do with it.

 

... Which is probably why this will unfortunately never happen. If there's no array requirement, then with the current design, there's nothing stopping someone from just making a new bootable usb each time (copying over the config files from the last one) and never ponying up. Then there'd be nothing prodding folks to actually pay for a license, in the same tangible way at least (imo).

+1

+1. Would like to be able to run VMs and Dockers from unassigned devices and/or cache without starting the array.

It would be great to use this, because I use an Unassigned device for my VM, there is no need for me to use the Array for my VM, I use the array for cache, nextcloud or other things.

+1 for opnSense

+1 here.

 

Daft that I have to stop my PfSense VM to do various things like change SMB settings, change array settings, etc.

 

VMs/dockers should be able to run irrespective of the state of the array.

Edited January 21, 20215 yr by Interstellar

1 hour ago, Interstellar said:

VMs/dockers should be able to run irrespective of the state of the array.

How would you handle VM or container use of array resources?

On 1/21/2021 at 3:26 PM, jonathanm said:

How would you handle VM or container use of array resources?

 

Surely the same way that you do now for a path that doesn't exist? 

1 minute ago, Interstellar said:

 

Surely the same way that you do now for a path that doesn't exist? 

A path that doesn't exist currently is created in RAM when written to, and if you try to read from it, returns file not found.

 

I was asking for a way forward, as in how to accomplish what you are asking for.

26 minutes ago, jonathanm said:

A path that doesn't exist currently is created in RAM when written to, and if you try to read from it, returns file not found.

 

I was asking for a way forward, as in how to accomplish what you are asking for.

 

Really? Seems a bit odd to create a dir/file in RAM rather than say "does not exist"?

 

Alternatively, given that as far as I'm aware there is a fixed list of potential paths to the array(s). Therefore something like if the path contains "/mnt/user" or "/mnt/diskX" or "/mnt/cache" then it is an array dependent VM/docker and will be skipped at initial docker/VM starts.

 

Then when you start the array you can see which dockers/VMs haven't been started due to having an array dependency and start them then.

 

Given only one of the 20 odd VMs/Dockers I currently run depend on the array (and is actually the only reason I have a "cache" drive (I don't use the array cache feature because of performance reasons) and some would benefit from not being array dependent, then its a fairly key feature for me going forward and actually the only one I can currently think of that I need from UnRAID.

 

Edited January 22, 20215 yr by Interstellar

Very much want this feature.

+1

 

I could definitely see this as a very useful feature.  I would prefer that it go to unassigned drives or extra pools when they're available.  I'd prefer not to put them on my cache drives.  As I add network necessary VMs like pihole, I hadn't really thought about all DNS services being down while my array was down.

Another +1 for this. If the VM/Docker Container has no paths mapped to the array, and runs on an unassigned device, allowing it to start with the array running and to continue when the array is stopped would be most useful.