Moving disks to a new system

[vitaprimo]

My server's been crashing every few days; out of nowhere it becomes unresponsive and I need to login to the management interface to hard reset it or go over there and push the power button until it goes power down.

 

Pushing the power button shortly though, makes the server beep hinting the unRAID's responsive enough to initiate shutdown but it doesn't complete it. I got videos of it for the first time now.

 

 

Sorry about the angle, it's a tight space.

 

Anyway, I think the server itself is failing, have been for a while. Reason I replaced it with a new one some time ago and I installed unRAID on it to see how it fared. Not good it seems.

 

I have another spare computer where I can move the disks to but it's a very different system: single CPU, non-hyperthreaded, non-ECC, non-Xeon family but still an Intel processor (Core i) and still with enough space (SATA ports) for the simple disk (1+1) setup I have on this one.

 

Could I move the disks from one server to another without rebuilding the array? If so, how's it done? I'm guessing taking the array offline and gracefully shutting it down to begin with but I don't know if it's going to be able to be imported.

 

Any help is greatly appreciated !

[itimpi]

 Normally it is as simple as shutting down the first server;  transferring the USB stick and disks to the new server; and then booting up the new server.  UnRAID is largely hardware agnostic and recognises disks by their serial number which is why it typically ‘just works’ after such a transfer.   This does rely on both systems having simple HBA type disk controllers so that the disk serial numbers are presented to Unraid in the same way on both systems.    Also if you have VMs with hardware pass through they will need redoing for the new server.
 

 

 

 

 

[Frank1940]

Yes,  You can run a basic server (a NAS) on just about any hardware.  VM's may present some challenges but most of the time, things can be made to work.

 

One thing you want to do is go to the Main page and print out that page with all of the disk assignments.  (You will probably never need it but having it will guarantee that!!!) 

 

[vitaprimo]

Thanks you both so much for your help.

 

I had been postponing adding disks and complexity to the system and it seems it worked for the best. I only have containers in it, actually they're the main reason I'm using (and like) unRAID so much, it's a great file management and sorting tool.

 

Assignment should be easy as I was still only working on 1 disk (plus 1 parity). Might be a good time to dust things off; the server room is a tiny closed air-conditioned (ductless) room but for some reason the servers still manage to get a lot of dust trapped in them. It's annoying but whatchugon'do.

 

Thanks again !

[vitaprimo]

After switching the disks to a new system I was pleasantly surprised indeed it came straight back up, I had to only change the network settings for the LACP bond. I was going to report a few days later everything went OK but the system crashed yet again.

 

Because of the processor core count going down from 16 (hyperthreaded though) to 4 (real, non-hyperthreaded) it was easier to see the processor was busier I digged a little and found it to be syncthing. I killed the container and rebooted the server and it went without crashes a little further.

 

It was now very obvious it's not a physical problem so I moved the disks back to the 16-core server with syncthing turned off and it lasted way longer and technically it didn't crash this time but it did stop responding overnight (I play old TV series to help me sleep; hardcore insomniac). I pushed the power button on the server as soon as I sit up (I slept in the server room floor for testing) and it shut down correctly, move the disks on last time to the 4-core and installed Windows Server 2019 on the 16-core server to move the data back to it preserving DHCP/DNS-vs-MAC and rack placement/labeling.

 

Moved the data at about 2Gbps on spinning disks which was rather impressive because moving the data to unRAID was insanely slow, I assume because of parity but Storage Spaces in Windows Server also does parity calculations and it was ready while I showered and blow-dried my hair.

 

Once in Windows, I found out that all of the software I used in containers was available for it and I got it all including syncthing and waited it out for crashes and there have been none. I hate Windows but gotta give credit where it's due, Windows is very hard to crash and as long as it doesn't connect to the internet to autof_ck itself up it's a very stable, very resilient OS. Also, if it's not connected, there's no worries about security. That doesn't mean it's not protected, it's domain-joined so all security is automatically applied plus a few tricks to hinder telemetry and Windows Update right from within itself and Active Directory is the one good product Microsoft produces–even if it's a ripoff like the rest of them.

 

Now for the bad, because I was on the bad server during the trial, I blamed the crashes on the server hardware plus a combination of factors, like power outages that occur every year around August because of the weather and restructuring of the data from different servers while keeping the hypervisors running (from iSCSI) and learning how to use unRAID. On the first days it had way more containers and ran for up to 15 days without crashing too, so that might not be it either.

 

 

 

I think the first logs still apply. I'm leaving unRAID for now to use something stable but since I now have a license I might give it another go in the future. I hope this helps developers or something.

 

 

Anyway... See you!

Edited September 20, 2019 by vitaprimo
Wrong image