3 hours ago, extrobe said:

I'd love to see partial-parity scheduling.

eg, do 25% each week, meaning a full scan is done every 4 weeks. Currently, doing a full scan takes nearly 25hours, so don't tend to run them unless I have had a dirty restart or something. Being able to phase it would allow users to keep parity in check

Probably more important is the ability to continue a parity check (or other long-running array operation) after a reboot sequence.   In fact if there was even a way to start such an operation starting at a defined offset (rather than always from the start) the parity check tuning -plugin could easily be enhanced to do the rest.

On 11/23/2019 at 9:20 PM, coolspot said:

 

Typically means version control, a copy of the file is stored in the metadata and can be restored at will. It's a feature of BTRFS and most major OSes like Windows can support them (i.e. right click, choose previous versions, etc.)

 

They're great for preventing ransomware because you can take snapshots frequently without incurring a lot of space with BTRFS and can roll back chances en masse very easily.

doesn't zfs has this feature, i voted for zfs, vm enhancements and trim support. but define vm enhancements? the only reason people use freenas is because of zfs, so unraid could kill 2 birds with 1 stone, and add zfs support as that has snapshot support.

On 11/23/2019 at 1:40 PM, vagrantprodigy said:

Snapshots (For VMs and shares) would be huge. Right now it absolutely stops me from being able to recommend this to even a small business.

I've been doing this for a long time now via command line with my important VM's.

 

First, my VM vdisk's are in the domains share, where I have created the individual VM directory as a btrfs subvolume instead of a normal directory, ie:

btrfs subv create /mnt/cache/domains/my-vm

results in:

/mnt/cache/domains/my-vm  <--- a btrfs subvolume

Then let vm-manager create vdisks in here normally and create your VM.

 

Next, when I want to take a snapshot I hibernate the VM (win10) or shut it down.  Then from host:

btrfs subv snapshot -r /mnt/cache/domains/my-vm /mnt/cache/domains/my-vm/backup

 

Of course you can name the snapshot anything, perhaps include a timestamp.  In my case, after taking this initial backup snapshot, a subsequent backup will do something like this:

btrfs subv snapshot -r /mnt/cache/domains/my-vm /mnt/cache/domains/my-vm/backup-new

 

Then I send the block differences to a backup directory on /mnt/disk1

btrfs send -p /mnt/cache/domains/myh-vm/backup /mnt/cache/domains/myh-vm/backup-new | pv | btrfs receive /mnt/disk1/Backup/domains/my-vm

and then delete backup and rename backup-new to backup.

 

What we want to do is add option in VM manager that says, "Create snapshot upon shut-down or hibernation" and then add a nice GUI to handle snapshots and backups.  I have found btrfs send/recv somewhat fragile which is one reason we haven't tackled this yet.

 

Maybe there's some interest in a blog post describing the process along with the script I use?

 

I've been doing this for a long time now via command line with my important VM's.
 
What we want to do is add option in VM manager that says, "Create snapshot upon shut-down or hibernation" and then add a nice GUI to handle snapshots and backups.  I have found btrfs send/recv somewhat fragile which is one reason we haven't tackled this yet.
 
Maybe there's some interest in a blog post describing the process along with the script I use?
 

There sure is interest in a good write up of the snapshot feature. Command line is ok.

What about snapshots of important data directories on the array? Is this the same?

Sent from my chisel, carved into granite

10 hours ago, limetech said:

I've been doing this for a long time now via command line with my important VM's.

I've been doing the same, for several years now, I snapshot the VMs daily with a script without shutting them down, but since this results in a crash consistent state, I also turn then all off once a week and run another script for offline snapshots, if I need to go back I try to use an offline snapshot but have restored to online snapshots some times without issues, and this way I have ore options, since it's not very convenient for me to shutdown the VMs every day.

 

10 hours ago, limetech said:

I have found btrfs send/recv somewhat fragile which is one reason we haven't tackled this yet.

Interesting, I've been using send/receive for all my backups, including disk by disk full server backups to another Unraid server, and it's been working fine for me.

 

On 11/25/2019 at 4:04 PM, limetech said:

I've been doing this for a long time now via command line with my important VM's.

 

First, my VM vdisk's are in the domains share, where I have created the individual VM directory as a btrfs subvolume instead of a normal directory, ie:

btrfs subv create /mnt/cache/domains/my-vm

results in:

/mnt/cache/domains/my-vm  <--- a btrfs subvolume

Then let vm-manager create vdisks in here normally and create your VM.

 

Next, when I want to take a snapshot I hibernate the VM (win10) or shut it down.  Then from host:

btrfs subv snapshot -r /mnt/cache/domains/my-vm /mnt/cache/domains/my-vm/backup

 

Of course you can name the snapshot anything, perhaps include a timestamp.  In my case, after taking this initial backup snapshot, a subsequent backup will do something like this:

btrfs subv snapshot -r /mnt/cache/domains/my-vm /mnt/cache/domains/my-vm/backup-new

 

Then I send the block differences to a backup directory on /mnt/disk1

btrfs send -p /mnt/cache/domains/myh-vm/backup /mnt/cache/domains/myh-vm/backup-new | pv | btrfs receive /mnt/disk1/Backup/domains/my-vm

and then delete backup and rename backup-new to backup.

 

What we want to do is add option in VM manager that says, "Create snapshot upon shut-down or hibernation" and then add a nice GUI to handle snapshots and backups.  I have found btrfs send/recv somewhat fragile which is one reason we haven't tackled this yet.

 

Maybe there's some interest in a blog post describing the process along with the script I use?

 

Why not use libvirt's built in snapshot? https://wiki.libvirt.org/page/Live-disk-backup-with-active-blockcommit

 

 

13 minutes ago, ionstorm66 said:

Why not use libvirt's built in snapshot? https://wiki.libvirt.org/page/Live-disk-backup-with-active-blockcommit

Wanted to play around with btrfs send/recv to save backups to another server.

Would the libvirt snapshotting work on xfs?

On 11/25/2019 at 1:04 PM, limetech said:

Maybe there's some interest in a blog post describing the process along with the script I use?

 

Yes, yesssss. 

8 minutes ago, BRiT said:

Would the libvirt snapshotting work on xfs?

it works as long as you are using qcow2/raw for the vm disks

I don't really know what most of those options are in the poll.

Probably the wrong place to ask, but I would really love a solid built in way to see what service/docker is spinning up my disks n whatnot (I use open files / file activity plugins, but my disks are always spinning)

I voted snapshots though I must say basic definitions of the features would have helped. The number one improvement everyone should still be asking for is a file explorer than isn't a half ass version of Krusader or Dolphin. It doesn't make sense this far into the project to not have made a good solid GUI based file explorer. Having FOLDER based version is probably all most people need. 

It's a server mostly for large cheap storage pools. You want features for data management, exploring data, observing data use, monitoring data changes. You want features for backup and file syncing all built in and if not robust at the very least fairly seamless and fool proof. Snapshot makes sense, expanding arrays/drives makes sense for SOME, but probably less ppl than anything like Version. 

 

The other BIG weakness of unraid is file sharing to insecure windows desktops. I think building basic client folder syncing in makes the most sense, but it all needs a well polished UI. I didn't buy unraid because I wanted to drop to command line or use Krusader. I bought it as an alternative to getting balls deep in Linux apps, command and scripts because that all takes TIME I don't want to spend.

That being the case when it comes time to share a file my natural response is to GET ER DONE and that means a public share or simple SMB password. That's ok, but it could be a lot better in this day and age and it could save untold bits of data from malicious corruption via those super insecure network shares, which are mostly not being used in a direct access fashion anyway. 

For all the people that just want to put and pull files from Unraid and browse and delete files from shares it makes good logistical sense as a product and consider the likely customer base to build that feature it. 

OR just look at the threads as to what questions come up the most and add features so the problems don't exist. ;) That's the easy way to do things. Solve the most common problems and call it progress, chances are you've done more good that way than experimental features which benefit small factions of users. 

 

NONE of those features seem like things mainstream users are likely to need other than simple easy to use versioning... if that's what snapshot would turn out to mean.  

 

 

 

 

On 11/23/2019 at 6:15 PM, unRate said:

My number 1 wish is better security https://forums.unraid.net/topic/80192-better-defaults/

My number one request as well. I would like Unraid to catch up to modern distros here and maybe we could even start recommending it to SMBes as an alternative to FreeNAS and OMV. 

 

PS I can definitely help with implementing a few of these. I just would like to use a better solution than running a script at boot to do it. 

Edited November 30, 20196 yr by xanvincent

I'd love to have an ability to limit max size of shares. The other day my dad scheduled 300 movies with Ombi and rendered my NAS to a halt when it reached 100% capacity. Of course I can set disk exclusions, but that's not a proper solution. 

9 hours ago, Arandomdood said:

I'd love to have an ability to limit max size of shares. The other day my dad scheduled 300 movies with Ombi and rendered my NAS to a halt when it reached 100% capacity. Of course I can set disk exclusions, but that's not a proper solution. 

Yes!  I would love to set max size for a particular share as well!  +++++1

On 11/25/2019 at 1:04 PM, limetech said:

I've been doing this for a long time now via command line [...]

 

Man pages, man pages, man pages please!

 

I get that the goal for unraid is to build functionality into the GUI, but as a linux-savvy user, I will not hesitate to use the command line. Having worked with countless versions of *nix commands over the decades, I want to be able to confirm the specific behavior of the versions installed in Unraid.

 

Thanks!

On 11/23/2019 at 2:30 PM, SpencerJ said:

The fine people over at the "unRAID Users & Help Group" on facebook are running a poll. Vote here, vote there and let us know:

What new features would you like to see the most in 2020? Please comment below to add your feedback. 

 

Note: You must request to join their group.

"Fine people" and "on facebook" do not compute :P

Ask for Chinese support, Chinese users really urgently need Chinese support

4 hours ago, djgizmo said:

"Fine people" and "on facebook" do not compute

Normally true- but they are a friendly group

So I'd really really like to see VM management made a little easier;

 

1. Non destructive editing

I'm getting really tired of copy / pasting XML fragments any time I made an edit in VM manager form view. Yes, I run OSX VMs that need some special stuff in XML, but it would be great to be able to change GPUs etc., without needed to paste all the custom XML back in.

 

2. Easier Passthorugh

We can define CPU pins and ACS override through the GUI, why can't we specify PCIe devices to hide? A simple checkbox beside an IOMMU group in the device listing that would drop it into the syslinux file would be a super time saver. (and make explaining device hiding to newbies a lot easier).

 

 

Edited December 22, 20196 yr by meep

2 hours ago, meep said:

A simple checkbox beside an IOMMU group in the device listing that would drop it into the syslinux file would be a super time saver. (and make explaining device hiding to newbies a lot easier).

This one is in the works.

 

Edit: actually, doesn't mess with syslinux.cfg file, instead it generates entries in config/vfio-pci.cfg

Edited December 22, 20196 yr by limetech
clarification

I would love to see the addition of cset shield in the VM Manager as an alternative to the defacto isolcpus requirement for separating compute resources for low latency.

Out of the options above, having ZFS would be the most appealing. As for items not on the list, I'd really like to see the inclusion of DVB-drivers into the default kernel.

Docker startup and shutdown dependencies (e.g so the database is shutdown second and started first), docker grouping and ZFS would be my votes.


Sent from my iPhone using Tapatalk

my 2 most requested is multiple arrays and snapshots.