Hello!
My Unraid server has suffered two power failures in the span of two days - there was some electrical work in my apartment complex and I coul have shut the server down, but I completely forgot. And as a result it got unceremoniously shut down twice.

 

I can't access the webgui. The login form shows up over https, but when I enter the credentials, I get a HTTP 500 error.
I am able to log in through ssh and monitor logs with 

tail -f /var/log/syslog

 

When I try to log in, I get:
 

Mar 26 16:05:46 theseus webgui: Successful login user root from 192.168.1.45

But still get a HTTP 500 in the browser.

 

I've tried disabling ssl, restarted nginx and php through ssh a couple times but that hasn't helped me at all.

I'm adding a few diagnostics files from today. I'd be grateful for any help.

tower-diagnostics-20250326-1505.zip tower-diagnostics-20250326-1510.zip tower-diagnostics-20250326-1528.zip tower-diagnostics-20250326-1548.zip

3 minutes ago, alucard87.pl said:

I coul have shut the server down, but I completely forgot

Get an UPS and configure Unraid to shutdown when it switches over.

 

Lots of this

Mar 26 15:34:51 theseus kernel: critical medium error, dev sda, sector 936927 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 2
Mar 26 15:34:51 theseus kernel: FAT-fs (sda1): Directory bread(block 934879) failed

then later

Mar 26 15:48:16 theseus root: Fix Common Problems: Error: Directory Bread Errors found ** Ignored

This isn't a warning that you should ignore.

 

Do you have a current backup of flash?

8 minutes ago, trurl said:

This isn't a warning that you should ignore.

You shouldn't ignore any FCP warning unless you can explain why it is OK to ignore for your specific configuration.

1. Hindsight is 20/20 and of course a UPS would have been a perfect method to avoid the situation. But I really don't feel like inviesting that much into a frankenstein server. I'd just be happy to get it running again with minimal possible data loss.

2. As far as I know (or remember), there were no drive errors before the power failures. I ran FCP somewhat regularly and did Array health checks monthly or thereabout, because the drives were bought second-hand with god knows how many work hours on them already. The server doesn't host any mission-critical data, but being able to recover it would be nice. It's a plain XFS JBOD.

I don't believe I have a flash backup that would be satisfactory.

 

2 minutes ago, alucard87.pl said:

no drive errors before the power failures. I ran FCP somewhat regularly and did Array health checks monthly or thereabout

Those errors are from your boot flash drive.

Just now, alucard87.pl said:

I don't believe I have a flash backup that would be satisfactory.

You must always have a current backup of the boot flash drive. It contains ALL of your configuration (ALL settings from webUI including disk assignments, docker templates, share settings, your license. ALL)

 

Unraid Connect plugin will keep a current backup of flash on Unraid Cloud. You can always manually download a zipped backup of flash from MAIN - Boot Device - flash - FLASH BACKUP. Any time any significant change is made in the webUI (your configuration) it needs to be backed up.

 

So now what to do...

 

The config folder from flash (if good enough) or flash backup are all you need to get going again with all of your configuration on a new install.

 

You should backup (in other words, copy) config folder from your boot flash to somewhere on your PC or wherever, format flash, and try to create a new install on that same flash drive. If that boots OK you can copy your config folder backup onto it.

 

If the new install doesn't boot, you might have to replace flash, but that also require transferring your license.

 

 

 

 

17 minutes ago, alucard87.pl said:

the drives were bought second-hand with god knows how many work hours on them already. The server doesn't host any mission-critical data, but being able to recover it would be nice. It's a plain XFS JBOD.

Your diagnostics showed no assigned disks so at least that much of your configuration couldn't be read from flash.

 

Did you have a parity drive?

 

19 minutes ago, alucard87.pl said:

second-hand with god knows how many work hours

SMART reports for all look fine with only a year or so Power_On_Hours

9 minutes ago, trurl said:

Your diagnostics showed no assigned disks so at least that much of your configuration couldn't be read from flash.

 

Did you have a parity drive?

Actually syslog does have something about that

Mar 26 15:34:55 theseus kernel: md: import_slot: 0 empty
Mar 26 15:34:55 theseus kernel: md: import disk1: (sdb) HGST_HDN724040ALE640_PK2334PEGNK5PT size: 3907018532 
Mar 26 15:34:55 theseus kernel: md: import disk2: (sdc) HGST_HDN724040ALE640_PK2338P4HYSBJC size: 3907018532 
Mar 26 15:34:55 theseus kernel: md: import disk3: (sdd) HGST_HDN724040ALE640_PK2338P4HYSN7C size: 3907018532 
Mar 26 15:34:55 theseus kernel: md: import disk4: (sde) HGST_HDN726040ALE614_K7GNWRDB size: 3907018532 
Mar 26 15:34:55 theseus kernel: md: import disk5: (nvme0n1) Patriot_M.2_P310_240GB_P310HHBB230822001791 size: 234430040 

So no parity.

 

Note on disk5. SSDs in the array cannot be trimmed. Usually these would be assigned to cache or some other pool.

Looks like license couldn't be read either.

Mar 26 15:34:55 theseus emhttpd: Unregistered Flash device error (ENOFLASH3)

Hopefully it can be copied or you have a copy.

@trurl Thanks, I managed to restore from a backup from myunraid.net (divine intervention, there was a backup from yesterday afternoon). Only some weird issues with restoring samba shares, but that might as well have been caused by Win11 f*ckery.

There's a number of issues (well, I like to call them project constraints lol) with the NAS as is. All of these will change with what I have planned, provided I scrounge up a decent budget for it. For now I'm learning the ropes (shares, dockers, vpn, etc.) and low-key trying to break it to see if I can fix it on my own. Though this crash was unexpected.

I'll get a better case and hardware. More and better storage. Then I'll set up ZFS and caching and all that jazz. UPS, VPN tunelling and access, etc.
For now I'm content with the lil' frankenstein chugging along under a table, presenting a public SMB share to my home network and a few services.

Sincere thanks for your explanations and patience.