[Solved] Can't connect to tower

[sonofdbn]

I was away for a few days, and did a clean shutdown on my tower, which is on unRAID 6.1.7. When I got home today I restarted the machine, which is usually headless, and found I couldn't connect to it from other PCs in the network. I attached a monitor and keyboard, and started the server again. I haven't done a cold boot in a long time, so can't remember what is meant to happen, but it did seem that unpacking bzroot took a longer time than I recalled.

 

Then the boot process reached "cpuload started" and after that I got an error "gzip: stdin" Input/output error"  followed by "tar: Unexpected EOF in archive" twice, then "tar: Error is not recoverable: exiting now".

 

The process continues, and I get the login prompt "Tower login:" Now, my memory is poor on this, but I thought that this should indicate that the server is up and running, but I couldn't connect to it from other PCs on the same (wired) network. I logged in as user root, but still couldn't see the server. I can ping "tower" with no problem, but I just can't get to the files on the server, or the web interface.

 

Also, my USB drive has a green light and my recollection is that this was always on while the server was running, but now there's no light. I don't know if this is relevant.

 

Any suggestions?

[Squid]

Try from the local keyboard after logging in:

 

diagnostics

 

And upload the file that will get saved onto the flash drive.

[John_M]

It would be worth checking the flash device for corruption by plugging it into another computer and checking it for file system errors.

 

[sonofdbn]

Thanks for the suggestions.

 

When I ran diagnostics I had a problem: I get a mkdir error, saying it can't create the /boot/logs directory. There is a message saying diagnostics file has been created, but when I transfer the flash device to another PC, there is no such file, and there's no /boot/logs directory.

 

I've checked the flash drive for system errors on a Windows 10 PC. Good old Windows tells me there's a problem with the flash drive and that I should scan it for errors, but when I do that Windows tells me there are no errors.

 

My suspicion is that the flash drive is failing. I've tried it on three PCs, and only one of the PCs "sees" it. I think I'm going to have to replace the drive.

[John_M]

It's a straightforward enough process. Good luck. https://lime-technology.com/replace-key/

 

[sonofdbn]

Hmm, not having much luck here. I copied the files on the current very old flash drive (Sandisk Cruzer Micro, from Lime-Tech in the early days) to a Sandisk Cruzer Blade, which I had done a guick format on (FAT32, labelled UNRAID). I then ran the make_boot as administrator. But the Blade doesn't seem to boot. I see the motherboard logo then just a flashing cursor. It doesn't get to the Memtest option screen.

 

I'm now trying to do a full format and retry the above, although I think that's unlikely to be the problem.

[Ashe]

Just check the bios is selecting the USB to boot, my gigabyte motherboard 'forgets' my USB if I take it out of the slot and I have to select boot from USB in the bios

[sonofdbn]

Thanks, Ashe. Yes, it was a BIOS issue. The Supermicro motherboard/BIOS saw the new flash drive as a USB hard disk, and placed it at the bottom of the list of hard disk devices for booting. I moved it to the top and now I'm up and running again - well as soon as the parity check is completed. Not sure why unRAID detected an unclean shutdown, but I haven't run one for a few months, so probably good to do.

 

Thanks to everyone for their suggestions.