I have a few questions around recovery anD backup.
I saw the thread about importing the raid1 pool so that is fine and i assume that will work from the web interface in future releases.
But are there any plans to add a feture that will create a backup config bundle so in the case when the usb stick you are running rockstor on dies you can do a fresh install, update the the latest version and import the config you backed up to your workstation reboot and all settings, config, shares are back.
Or will support be added to allow mirrored system pool on usb sticks or hdd?
That’s what I expected. I have an instance here running on sandisk cruzer and it is slow. The general usb slowness and unreliability is expected. I suggest turning off Metric Collection in dashboard unless you really need it. This will reduce writes on the usb stick.
Until that issue is fixed, there are a bunch of writes that can’t be easily avoided. This behavior is also usb unfriendly.
One more thing to know is that say your installed the OS on one of your 4 drives, a pool is automatically created on that disk and you can add Shares to it. So you can put data on it besides the operating system. However, as you may know already, there’s not redundancy on that disk.
I have explored that as an option but I want my data to be redundant as the purpose of this server is for backup data and the focus in on data integrity.
The way I have my other server setup is to have a boot & root partition from each data drive in a mirror with mdadm, a swap partition and then the remainder as a partition in the data array in raid 5.
I was hoping with rockstor to have a similar setup.
mdadm mirrored boot, swap partition and then the remainder in the btrfs pool for data.
Start with two drives then add more as required two at a time and convert pool to raid10.
Raid 5 would be out as I dont think you can boot from btrfs raid5 yet.
I have made some of the changes suggested and it is running well now not as well as when i had it on an ssd but i have another free sata port so that is a win.
I have followed the raid1 thread and re imported my raid1 pool and recreated my shares.
All sweet so I am happy that I can recover after an os drive failure it would be good to have that documented and the ability to import a pool via the gui would be great and then discover any sub pools recreating the samba and nfs shares is no big deal.
@sirhcjw a few days ago there was a large improvement in this area, ie see https://github.com/rockstor/rockstor-core/issues/534 a quick quote from @suman “… Repeat installations also get a nifty feature with this merge. If Rockstor is installed over an existing or broken install due to a root disk failure etc… all preexisting data in the form of pools, shares and snapshots are automatically imported. …” so that’s nice.
There is more info on these pretty key improvements in that issue if anyone is interested.
A nice effect of this work is “… that drives can be moved to another Rockstor machine where data can be imported from the UI with a few clicks, facilitating a reasonable recovery option.”
N.B. There’s a very recent issue opened that relates to certain imports:-
"While importing a BTRFS that was created outside of Rockstor, there may not be a qgroup which prevents new Shares from being created."
So that’s worth keeping an eye on:- https://github.com/rockstor/rockstor-core/issues/687
Linking here as relevant to the pool imports feature I was just discussing.