Unraid won't boot after power failure, USB drive seems fine

[FlowerPower]

SOLVED: unrelated to power loss. It is a BIOS issue that prevents USB boot with more than 11 drives installed. Solution: the solution to unraid not booting on a SuperMicro H8DGU(-F) motherboard, BIOS version 3.5c, with more than 11 drives is "go into the LSI configuration utility and change the adapter control from BIOS and OS to OS Only".

 

 

I tried to boot my server this morning and all I get is a blinking cursor post-BIOS (monitor connected to the server). I checked boot order in BIOS, tried putting the USB drive in a different slot, nothing. I put the USB drive in my Windows box and it appears to be fine. I was able to copy all files as a backup. How do I get this thing booting again? It's not even starting to boot as far as I can tell.

What happened prior:

 

I just finished the initial setup of my server, and was transferring some files. The power went out, but the UPS appeared to do it's job and shut the server down after five minutes. The UPS didn't power off as I had it set to, so I shut it down manually. When the power came back on, I powered up the UPS, the server started and then...didn't boot.

A number of changes had been made since previous shutdown and reboot. Three data drives added, two cache drives added, all shares except my media library set to cache only, Community Applications installed, Plex installed, and the UPS was set up. Everything was working perfectly, no warnings or errors.

Edited August 18, 2020 by FlowerPower

[trurl]

The config folder on flash contains everything needed to get going just as before from a new install. Since you say you copied that you should be good.

 

18 minutes ago, FlowerPower said:

I put the USB drive in my Windows box

Did you do checkdisk on it?

 

[FlowerPower]

Yes, both a quick scan and /f, no errors found. I assume my next step is to rebuild the USB drive and then copy over the old /config directory.

[FlowerPower]

Right, so  I rebuilt the USB drive with the flash creator, copied over config--no change in behavior. Is there something in my config that could cause the boot process to not even start? I get no screen output whatsoever.

[Frank1940]

2 hours ago, FlowerPower said:

all I get is a blinking cursor post-BIOS (monitor connected to the server).

I would take this to mean that the BIOS has decided to use the USB drive as the boot drive and it did find something which 'looked' like an OS.  (Otherwise, the BIOS would have complained but not finding a boot drive.)

 

What I would suggest at this point to build an all new USB boot drive using a new USB flash drive.  Plugin it to the server and see if that boots.  IF it does shutdown immediately.  Now copy the config folder from the backup and see if it boots.  (It will eventually complain about the .key not being valid but lets deal with that problem later...)

Edited August 18, 2020 by Frank1940

[FlowerPower]

I finally got an error message. I took all the boot devices out of boot priority in BIOS except the USB drive and removable devices. If I do that, I get the below, with both my original USB drive and a brand new one. If I add back in the other devices, it goes back to the blinking cursor.

 

It seems the server has stopped recognizing these as USB drives as bootable, although no changes should have happened in BIOS. What are the key BIOS settings to check for unraid? I am running a SuperMicro 36 bay server, a couple generations old. Motherboard is an H8DGU-F. I looked through every BIOS setting and don't see anything obvious. This is maddening. It worked perfectly through the whole trial period. Now that I'm licensed and have all my drives migrated, dead in the water.

 

[FlowerPower]

Well shit. On a hunch I pulled one the drives, which was added to the array but had no data on it yet. Sure enough, up pops the unraid boot screen.

So I can boot with 11 drives installed, but not 12. I read about a similar issue with a different motherboard, where USB was being treated like a drive, and it broke once you hit 12. Even though USB is definitely a separate boot option in this BIOS, I seem to have a similar problem.

 

This, needless to say, rather defeats the point of the 36 bay server. Any idea how I can get around it? MB s SuperMicro H8DGU(-F) version 3.5c, which appears to be the latest.

[Frank1940]

I would be looking at the BIOS.  Read the BIOS manual.  Someone might be using the same MB as you for Unraid.   But might take some time.  I would Google the problem and waht the wider universe has be say about boot settings.  .

[Frank1940]

7 minutes ago, FlowerPower said:

Well shit. On a hunch I pulled one the drives, which was added to the array but had no data on it yet. Sure enough, up pops the unraid boot screen.

So I can boot with 11 drives installed, but not 12. I read about a similar issue with a different motherboard, where USB was being treated like a drive, and it broke once you hit 12. Even though USB is definitely a separate boot option in this BIOS, I seem to have a similar problem.

 

I can remember seeing this a few years back too.  Can't remember what the solution was-- if there was one...  However, I would start by Googling with   unraid.net   as one of the search terms.  (Google does a much better of searching the Unraid forum than the forum's search tool!)

[FlowerPower]

Indeed it seems to be, and you were just solving the same problem a month ago. Looks like the same setting fixed it. 

 

For the next person who searches, the solution to unraid not booting on a SuperMicro H8DGU(-F) motherboard, BIOS version 3.5c, with more than 11 drives is "go into the LSI configuration utility and change the adapter control from BIOS and OS to OS Only"

 

[iStupid]

On 8/18/2020 at 5:44 PM, FlowerPower said:

Indeed it seems to be, and you were just solving the same problem a month ago. Looks like the same setting fixed it. 

 

For the next person who searches, the solution to unraid not booting on a SuperMicro H8DGU(-F) motherboard, BIOS version 3.5c, with more than 11 drives is "go into the LSI configuration utility and change the adapter control from BIOS and OS to OS Only"

 

Thank you! I had the exact same issue with not being able to boot after power outage and I can confirm this solved my problem as well - FINALLY. Had to push Control+C at boot to enter LSI Configuration in order to change to OS only.

[srtj]

My unraid server is behaving in the exact same way, but I have many less drives than @FlowerPower... I have 3 data drives, 1 parity, 1 cache for a total of 5.

 

My server has behaved this way one time before ~4mo ago, when I lost power and power did not come back on before the UPS was out of juice. In this instance, I bought a new flash drive, recreated my boot drive, and then requested a new license / GUID and all was well. The USB drive is being recognized by Windows and I can manipulate folders and data on it, but when plugged into my unraid server as the boot drive, it notifies me that no boot media was found.

 

Has anyone figured out why this would happen with less than 12 drives installed? I have some critical scheduled events (plant feedings) being managed by scripts in HomeAssistant that can't rely on me having to be here to recreate a flash drive.

 

The other issue that seems to be happening is that my server doesn't do a clean shutdown when the UPS is running out of battery... looking for pointers on this as well if anyone has any advice.
 

Thanks in advance.