I had to reboot the server due to Plex issues and upon reboot, I get a "ping: unknown host github.com" on the console.  The GUI takes several minutes to load or to move between tabs.  No shares are accessible via SMB.  I can't even shut down the server without pressing the power button for 4-seconds which results in an unclean shutdown and parity check on reboot.  A normal shutdown attempt says "cannot execute /sbin/init"

 

A thread from last year on this error suggests I change the DNS to 8.8.8.8 and 8.8.4.4 (google).  I can't through the GUI.  Seems to be related to plugins.  Should I reboot in safe mode and see if GUI is more responsive and change the network settings as a test?  Can I shutdown and edit network.cfg directly on the flash drive.

 

RIght now nothing is usable.  Yeah, I want to get it fixed, but, I would also liked to know what caused this on reboot.

 

I already ran chkdsk on the flash drive in case there was an issue there.

 

UPDATE:

Tried editing network.cfg on the flash and noticed the DNS servers are already set to the Google DNS

 

UPDATE 2:

Tried to boot in safe mode without GUI, GUI login came up anyway but the GUI never appeared; just said waiting for localhost for a couple of minutes.  I just have a blank browser screen on the console.

Edited August 24, 20178 yr by Hoopster

Diagnostics attached

medianas-diagnostics-20170823-2144.zip

Use a USB 2.0 port for the flash drive, USB 3.0 is not recommended.

i had the flash drive in a USB 2.0 port when the problems began.  I only switched to a usb 3.0 port when I noticed some usb errors in the syslog. I'll switch it back. However, I cannot shutdown the server at all without the power button because of the "cannot execute /sbin/init" errors.  The unclean shutdown starts a parity check on reboot and it takes several minutes to do anything through the GUI (like cancel the parity check) because everything is slow as molasses.

Edited August 24, 20178 yr by Hoopster

OK, did a hard shutdown of server; formatted a new flash disk; copied the contents of the old flash drive to new flash; made bootable; replaced bzimage, bzroot and bzroot-gui with the file from a fresh download from Limetech and rebooted.  I still get the "ping:unknown host github.com" error on the console, the GUI remains terribly slow and I cannot shutdown the server because of "cannot execute /sbin/init" without doing a hard shutdown with the power button. 

 

Gotta run for about 45 minutes, willing to try any and all suggestions when I return.

Edited August 24, 20178 yr by Hoopster

Post new diagnostics.

New diagnostics attached.  I also switched LAN cables.  I have another unRAID server on the same LAN, my backup server, running 6.4 rc7a, it is running fine.

 

I did notice also that, since the server having the problems is set as the SMB Local Master, none of the other Windows computers on the LAN show any other computers in the Network in Windows explorer.  They can ping each other.  In Windows Explorer, the backup server is accessible by name or IP address.  The main server with the problem is not accessible either by name or IP address. 

 

 

medianas-diagnostics-20170824-0414.zip

You're using a static IP, change to DHCP to see if you get an address, so you'll know it's working as it should.

I tried to change to DHCP through the GUI, but, besides being slow, I was unable to disable the VM Manager so I could change the network settings.  The dropdown to select No appeared but there was no Apply or Done button so VM Manager remained enabled.  I had no problems disabling docker.  It took me this long to reply because of how slow the GUI is.

 

In the end I had to shutdown the server manually again (i've probably cancelled parity check for the 10th time now) and edit the network.cfg file on my desktop.  I set it to DHCP and the server was assigned the address of 192.168.1.36 when booted and I can ping it at that address.  Still not accessible in Windows; looks like a Windows permissions issue.

 

The old familiar errors are still there (can't ping github.com) and can't execute /sbin/init on an attempted shutdown.

I would completely redo the flashdrive, copy only your key file and super.dat from current one.

Edited August 24, 20178 yr by johnnie.black

1 minute ago, johnnie.black said:

I would completely redo the flashdrive, copy only your key file and super.dat from current one.

 

So, do a new config? Don't copy any config files at all and redo all shares, VMs, Dockers, etc?

You don't need to do a new config, backup your current flash, format, unzip a a new unRAID install and only copy from the backup the key file and super.dat.

 

Super.dat is where the disk assignments are stored, so the array will be configured, and if all works you then just need to enable docker and point to the old image, and enable VMs and point to the old libvirt.img.

OK, thanks, got it.  I'll let you know what happens.

One thing I forgot, when configuring docker make sure you also point to the old appdata location.

32 minutes ago, johnnie.black said:

One thing I forgot, when configuring docker make sure you also point to the old appdata location.

 

Yeah, I have had to reconfigure dockers before, so I knew I would need to do that.

 

I low-level formatted (deselected quick format) the flash drive and did a clean install copying only super.dat and the key file.  So far, so good.    No ping github errors and after configuring network (back to static IP) and server identification I rebooted with no errors.  So far, so good.  Something very important on the flash drive must have become corrupted.

 

Frankly, I had considered that I might need to start over with the flash drive.  I had to do that a couple of months ago as well.  That's the good thing about having a card reader flash drive.  I can just swap in a new card and keep the old one in case I need anything from it.  Thanks for your help thus far, I was hoping to avoid the new flash route, but, who knows what was causing those errors, it likely would have been very difficult to track down.

1 hour ago, Hoopster said:

who knows what was causing those errors, it likely would have been very difficult to track down.

 

If you really want to find out you could replace the current flash files one at a time from the old flash backup, it will probably be some file in the config folder, if you do it backup current working flash first so you don't need to repeat the process when you find the culprit.

Well now on to the next issue.  I pointed dockers to the existing image file and all were there.  At first the WebUI opened on each one of them (except DelugeVPN).  I even played a couple of movies on Plex through the WebUI and all seemed well.  Now the only Docker WebUIs that will consistently open are Handbrake and Dolphin which are XRDP based.  Crashplan, DelugeVPN, Plex and Syncthing will not open; sometime Crashplan does, but, the other three will not.  Don't know if I have a corrupt docker.img file but, I will delete it and try reinstalling dockers.

 

The behavior started after a BTRFS scrub on the docker image after which it jumped from 3GB used to 10GB used. I have no idea if that has anything to do with the issue, it was just the last thing I had done before the WebUI problem appeared.

21 minutes ago, Hoopster said:

The behavior started after a BTRFS scrub on the docker image after which it jumped from 3GB used to 10GB used.

IMO, its pretty much pointless to ever do a scrub or any sort of file system repair on the docker image.  Easier and less aggravating to always simply delete it and then re-add the previously installed apps.  5 minutes 

29 minutes ago, Squid said:

IMO, its pretty much pointless to ever do a scrub or any sort of file system repair on the docker image.  Easier and less aggravating to always simply delete it and then re-add the previously installed apps.  5 minutes 

Yes, probably true.  And that's what I'll do in the future.