[SOLVED] No access to webgui

[dheg]

The system was working fine until yesterday. I had a power outage but the UPS did its job and as far as I know the shutdown was clean.

I started the server this morning and I cannot  access the webgui. The rest seems to be working fine.

• I can SSH into the system.
• All dockers are running fine, I can access their webguis
• VM running as well (started from the command line)

 

I reboted and same behaviour, no error message and everything seems fine.

 

Any ideas?

 

 

 

gaia-diagnostics-20210725-2232.zip

Edited July 27, 2021 by dheg

[dheg]

Anyone?

[Praetorr]

I had the exact same problem. In the end, I re-imaged my flash drive from a backup. That still didn't work. Then after rebooting the server around 10 total times, it eventually started working randomly. That was about 2 months ago, and it seems to be continuing to function... 

 

I don't really have anything too helpful to add, sadly. My theory is that something just doesn't work right when the server reboots from a UPS-driven shutdown. As I said, what happened to you was pretty much identical to my experience.

[dja]

Did you get this sorted?

I'm seeing this repeated near the end of logging quite a bit.  (Line 3231 in syslog)
Something to look at possibly. 
 

Jul 25 22:06:08 gaia root: Starting Nginx server daemon...
Jul 25 22:06:08 gaia nginx: 2021/07/25 22:06:08 [emerg] 19819#19819: bind() to 0.0.0.0:443 failed (98: Address already in use)
Jul 25 22:06:08 gaia nginx: 2021/07/25 22:06:08 [emerg] 19819#19819: bind() to [::]:443 failed (98: Address already in use)
Jul 25 22:06:08 gaia root: nginx: [emerg] bind() to 0.0.0.0:443 failed (98: Address already in use)
Jul 25 22:06:08 gaia root: nginx: [emerg] bind() to [::]:443 failed (98: Address already in use)

 

Edited July 27, 2021 by dja

[dheg]

3 hours ago, dja said:

Did you get this sorted?

I'm seeing this repeated near the end of logging quite a bit.  (Line 3231 in syslog)
Something to look at possibly. 
 

Jul 25 22:06:08 gaia root: Starting Nginx server daemon...
Jul 25 22:06:08 gaia nginx: 2021/07/25 22:06:08 [emerg] 19819#19819: bind() to 0.0.0.0:443 failed (98: Address already in use)
Jul 25 22:06:08 gaia nginx: 2021/07/25 22:06:08 [emerg] 19819#19819: bind() to [::]:443 failed (98: Address already in use)
Jul 25 22:06:08 gaia root: nginx: [emerg] bind() to 0.0.0.0:443 failed (98: Address already in use)
Jul 25 22:06:08 gaia root: nginx: [emerg] bind() to [::]:443 failed (98: Address already in use)

 

 

Thanks for this, it's not yet sorted, I was hoping for some clues.

The only thing I can think of is that I have assigned those ports to docker.

Do you know how can I check docker from the command line?

[dja]

4 minutes ago, dheg said:

 

Thanks for this, it's not yet sorted, I was hoping for some clues.

The only thing I can think of is that I have assigned those ports to docker.

Do you know how can I check docker from the command line?

Well, I would disable docker and reboot. See if that clears it up. You might try this to get GUI and disable the service. 
/etc/rc.d/rc.docker stop

You could then do-
/etc/rc.d/rc.nginx-fpm restart
/etc/rc.d/rc.php-fpm restart

See if your GUI comes back, disable or otherwise investigate.  *Disclaimer, not a pro. Don't think this will hurt, anything- but I'm putting it out there!

[dheg]

  

3 hours ago, dja said:

Well, I would disable docker and reboot. See if that clears it up. You might try this to get GUI and disable the service. 
/etc/rc.d/rc.docker stop

You could then do-
/etc/rc.d/rc.nginx-fpm restart
/etc/rc.d/rc.php-fpm restart

See if your GUI comes back, disable or otherwise investigate.  *Disclaimer, not a pro. Don't think this will hurt, anything- but I'm putting it out there!

 

I fixed it, one of the dockers (Heimdall) was using port 443, you put me in the right path!

Instead of messing with the command line, I tried first to start in safe mode, the webgui loaded without issues.

I changed the docker port to 443 and rebooted, everything is fine now.

 

Thanks!

[dja]

15 minutes ago, dheg said:

  

 

I fixed it, one of the dockers (Heimdall) was using port 443, you put me in the right path!

Instead of messing with the command line, I tried first to start in safe mode, the webgui loaded without issues.

I changed the docker port to 443 and rebooted, everything is fine now.

 

Thanks!

Nice! Glad it worked! 

I love Unraid, but if I *had* to complain about anything it might be config things like this that can creep up when you accidentally set a setting the wrong way and it conflicts and breaks something else. Maybe docker can get some error/conflict checking when assigning network ports/IPs in the future.  I know...we should be more careful!

[ChatNoir]

9 hours ago, dja said:

I love Unraid, but if I *had* to complain about anything it might be config things like this that can creep up when you accidentally set a setting the wrong way and it conflicts and breaks something else. Maybe docker can get some error/conflict checking when assigning network ports/IPs in the future.  I know...we should be more careful!

You can use the same ports on several dockers as long as they do not run at the same time.

Preventing this might cause issues for some users.

[JonathanM]

2 hours ago, ChatNoir said:

You can use the same ports on several dockers as long as they do not run at the same time.

Preventing this might cause issues for some users.

Yes, I exploit this for a bit of security through obscurity. I have multiple file management containers with different levels of array access, all running on the same port. The least dangerous container is set to auto start. None of the more dangerous containers will start while the neutered one is running, so gaining full access with Krusader means shutting down one container and manually starting the other. They all answer on the same port, so it all looks the same from the client end.

[dja]

2 hours ago, jonathanm said:

Yes, I exploit this for a bit of security through obscurity. I have multiple file management containers with different levels of array access, all running on the same port. The least dangerous container is set to auto start. None of the more dangerous containers will start while the neutered one is running, so gaining full access with Krusader means shutting down one container and manually starting the other. They all answer on the same port, so it all looks the same from the client end.

That makes sense, and I get it- but for IPs it should be easy to throw a ping out before applying the change and if a reply is received- warn the user that it may be in use? (Do you want to continue?) 

Edited July 28, 2021 by dja

[ChatNoir]

You can (and should ?) "Show docker allocations" when adding a new docker.

 

Could Unraid and/or CA add a dialog ?

Maybe, but I think that it is not as trivial as it sounds.

[Squid]

When installing or editing a container, if there are any conflicts (or if another error happens which prevents starting), it will appear in the docker run command which is listed.

 

If just starting an already installed container and due to a conflict / error it won't start you'll see "Server Execution Error", at which point you can easily see that actual issue by editing, make a change, undo the change and then hit apply

 

As an aside, 6.10 when changing any of the ports in the UI does a check to see if the port is already in use and won't let you apply the changes.