Segfaults, broken GUI

[ventrue]

Just for the record: Still new here, still trialling Unraid 🙂

 

Can someone help me solve this?

 

I'm having some issues that are probably related:

• Broken Web GUI: Parts not loading, Bad Gateway or PHP errors like "Fatal error: Allowed memory size of 134217728 bytes exhausted (tried to allocate 4294967568 bytes) in [varying scripts]". This gets increasingly worse until the GUI become unusable.
• Lots of segfaults in the log, for multiple applications (bash, php, ngnix, libc)
• Bad transfer performance. This might be unrelated and simply a hardware or network limitation though, not sure. The other day, I got 10 MB/s writing on the NAS, today I reformatted, used XFS instead of btrfs this time, and got 30 MB/s.

 

What I've tried so far:

• Closing the console (with htop in it) and log seem to at least temporarily restore GUI functionality
• Problems can reoccur right after rebooting, not not necessarily runtime related
• Problems seem to suddenly appear or get much worse when files are transferred to the NAS
• Two passes of Memtest haven't shown any errors. I'm running more now.
• Completely reinstalling Unraid
• Docker and VMs are disabled
• User Applications, unBalance and Dynamix sleep are installed

tower-diagnostics-20210627-1800.zip

[Squid]

First thing you want to do is sort out the 

Jun 27 08:27:32 Tower emhttpd: Unregistered - flash device blacklisted (EBLACKLISTED2)

if only so that those errors get removed from the syslog (and there's a ton of them)

 

This *may* also contribute to your problems, since the Fatal error: Allowed memory size of 134217728... refers to the space being taken up by the logging being exceeded

[ventrue]

Yeah, I did notice those blacklist errors as well. Not sure what happened there. They seemed pretty inconsequential but to make sure, I got out another flash drive and literally started from scratch with a fresh install, a new trial key and no backup import whatsoever.

 

Same issues as before: The segfaults pop up basically with every click in the half-working GUI as long as a file is being sent to the NAS. Cancelling the transfer seems to solve the issue.

 

I'd also like to point out this line in the log:

Quote

Jun 27 12:54:14 Tower emhttpd: Unregistered - flash device blacklisted (EBLACKLISTED2)

I believe this is the moment that blacklist error starts appearing on this new flash as well. Would surely be interesting to know what this means, but the other issue appeared first, so maybe they aren't related.

 

Edit: Checking back after a while, the errors actually do continue even without accessing the NAS. It's still parity-syncing, maybe it'll get better when that is finished. Edit 2: No it doesn't.

 

tower-diagnostics-20210627-2201.zip

Edited June 27, 2021 by ventrue

[ventrue]

Any more ideas, anyone? Any further details I can provide to help solve this?

[trurl]

Your flash drive must have a unique GUID. If the GUID doesn't appear to be random data then it is probably not unique.

 

 

 

[ventrue]

Looks fairly random to me on both drives I've tried (one PNY, one SanDisk). I agree that after a while, a pattern seems to develop where each segfault is accompanied by a blacklist-error. This is different in the beginning, however. As I said, the segfaults appeared first and were also significantly more numerous at first.

 

And in any case, why would I be issued a trial key for a blocked GUID to begin with and why would it not stay unusable permanently if it was indeed blocked, rather than immediately revert to a usable state every single time?

[trurl]

Probably your flash drive has disconnected. Boot from USB2 port.

[ventrue]

It already is in a USB2 port. I could try USB1, if your idea is to just use any other type of port? It's fairly old hardware, so I don't have a whole lot of options here.

 

The log doesn't show any disconnects though. The drive (or rather the share) also seem accessible all the time, although this is certainly not a bombproof way to find intermittent problems.

[trurl]

RAM is also a suspect but you say you have already tested.

[ventrue]

Yes, Memtest didn't find anything at least. Could always be some freak hardware bug, I suppose.

 

Other USB ports don't show different behaviour. It always works fine at first, then segfaults show up (seemingly in relation to a file transfer, haven't yet seen it before that) and only after that the blacklist errors also appear.

 

I've had the storage running for over a day now with constant access and even though these errors keep popping up in the log, the storage itself seems to work fine. If the GUI wasn't affected, I wouldn't even notice that something's wrong.

 

I've also sorted out the speed issue, it's definitely limited by the network. Had to reposition some Wifi stuff.