[Support] - Unifi-Controller: Unifi. Unraid. Reborn.

[yverde]

Hi,

I'm trying to restore a backup from the official desktop controller app.

When I use this container instead my AP shows as adopting then goes to Offline then back to adopting, but never succeed.

 

 

 

The AP is not offline though as all my devices are not connected to it. 

I'm using the new version 8.0.28

 

Should I be able to restore from a backup made in the official controller?

 

Thanks!

 

[PeteAsking]

38 minutes ago, yverde said:

Hi,

I'm trying to restore a backup from the official desktop controller app.

When I use this container instead my AP shows as adopting then goes to Offline then back to adopting, but never succeed.

 

 

 

The AP is not offline though as all my devices are not connected to it. 

I'm using the new version 8.0.28

 

Should I be able to restore from a backup made in the official controller?

 

Thanks!

 

Yes you should be able to. 
 

However, as noted:

 

“8.0.28-unraid (Version skipped, did not pass testing).”

 

As you are using an unsupported tag, any issues are up to you to resolve. We do not support tags that are unsanctioned and you are told to skip. You can use whatever tag you want, but if it falls outside what is sanctioned then its on you to resolve any problems. 

Edited February 23 by PeteAsking

[yverde]

10 hours ago, PeteAsking said:

Yes you should be able to. 
 

However, as noted:

 

“8.0.28-unraid (Version skipped, did not pass testing).”

 

As you are using an unsupported tag, any issues are up to you to resolve. We do not support tags that are unsanctioned and you are told to skip. You can use whatever tag you want, but if it falls outside what is sanctioned then its on you to resolve any problems. 

Ah I misunderstood, I thought it was unsupported until they release it here https://hub.docker.com/r/11notes/unifi/tags

But it is actually completely skipped I see.

 

Thanks for the clarification!

[PeteAsking]

15 minutes ago, yverde said:

Ah I misunderstood, I thought it was unsupported until they release it here https://hub.docker.com/r/11notes/unifi/tags

But it is actually completely skipped I see.

 

Thanks for the clarification!

It was discussed and understood that some critical feature might be required for someone in a new version and we cant prevent someone using a new version altogether if it is absolutely critical for their operations they do so, but at the same time we can only really support images that we vetted to actually work properly since this is the solution that covers 99% of people and ensures minimal breakage. 

[bmartino1]

Wanted to thank you PeteAsking for your continued support and maintaining of the docker.

 

Also wanted to throw this out to the community for those who would want to maintain their own instead of waiting for a tag release.

 

use a LXC container:

 

Edited March 3 by bmartino1

[NeuroDoc]

First, thanks for creating this!

 

Second, I'm having problems getting the app to run.  I'm using 11notes/unifi:8.0.26-unraid, and everything seems to start-up correctly (no warnings during docker build).  However, I'm unable to get the UI to load.  Looking at the log, the only error is:

 

WARN Unable to load properties from '/usr/lib/unifi/data/system.properties' - /usr/lib/unifi/data/system.properties (No such file or directory)". 

 

However, when I go into the console, the directory and file are there.

 

Any suggestions?

[PeteAsking]

4 minutes ago, NeuroDoc said:

First, thanks for creating this!

 

Second, I'm having problems getting the app to run.  I'm using 11notes/unifi:8.0.26-unraid, and everything seems to start-up correctly (no warnings during docker build).  However, I'm unable to get the UI to load.  Looking at the log, the only error is:

 

WARN Unable to load properties from '/usr/lib/unifi/data/system.properties' - /usr/lib/unifi/data/system.properties (No such file or directory)". 

 

However, when I go into the console, the directory and file are there.

 

Any suggestions?

Are you able to put the address in the url bar manually eg: https://<ipyouset>:8443 ?

[NeuroDoc]

4 minutes ago, PeteAsking said:

Are you able to put the address in the url bar manually eg: https://<ipyouset>:8443 ?

 

Yes, but I still get "connection refused".

[PeteAsking]

58 minutes ago, NeuroDoc said:

 

Yes, but I still get "connection refused".

Strange. What if you stop all your dockers then just start this one? Maybe one of the other dockers has consumed that port. 

Edited March 4 by PeteAsking

[NeuroDoc]

10 hours ago, PeteAsking said:

Strange. What if you stop all your dockers then just start this one? Maybe one of the other dockers has consumed that port. 

Ok, this seemed to work now.  I was also able to restart my other containers, and it's still working.

 

My only concern is that I had to set this container up on the bridge network.  For the old UniFi controller, I set it up under br0 and assigned it a unique IP.  I did this to maintain the AP adoption during software updates.  Going back to the bridge network now, I'm concerned that I will have to go through the adopt process each time the software updates...

[wgstarks]

6 hours ago, NeuroDoc said:

Ok, this seemed to work now.  I was also able to restart my other containers, and it's still working.

 

My only concern is that I had to set this container up on the bridge network.  For the old UniFi controller, I set it up under br0 and assigned it a unique IP.  I did this to maintain the AP adoption during software updates.  Going back to the bridge network now, I'm concerned that I will have to go through the adopt process each time the software updates...

Do you have Inform Host set to the IP of your controller?

 

I’ve never had to adopt my AP’s more than once.

[PeteAsking]

8 hours ago, NeuroDoc said:

Ok, this seemed to work now.  I was also able to restart my other containers, and it's still working.

 

My only concern is that I had to set this container up on the bridge network.  For the old UniFi controller, I set it up under br0 and assigned it a unique IP.  I did this to maintain the AP adoption during software updates.  Going back to the bridge network now, I'm concerned that I will have to go through the adopt process each time the software updates...

It sounds like one of the other dockers is using that port and if it starts first gets it. You will likely continue to have sporadic problems until you identify the conflicting docker. 
 

So far users found plex and jellyfin dockers both try to steal that port but there may be others. 

Edited March 5 by PeteAsking

[Elmojo]

11 minutes ago, PeteAsking said:

My only concern is that I had to set this container up on the bridge network.  For the old UniFi controller, I set it up under br0 and assigned it a unique IP

This is how I had to set mine up, for this very reason. Another docker was using port 8080.

If you want to know which it is, just open the Unifi template, scroll to the bottom, and expand the "show docker allocations" section.

It'll show you everything that is using ports, and which ones.

[JonathanM]

2 hours ago, Elmojo said:

It'll show you everything that is using ports, and which ones.

It shows bridge mode allocations.

Host mode containers open a whole can of worms, they have free reign to use whatever ports they want.

 

Don't use host if you can at all help it.

 

Do whatever it takes to figure out what ports are needed and use bridge mode.

[Elmojo]

14 minutes ago, JonathanM said:

It shows bridge mode allocations.

Host mode containers open a whole can of worms, they have free reign to use whatever ports they want.

How about that, I learned something new!

Thanks @JonathanM 

[Elmojo]

20 minutes ago, JonathanM said:

Don't use host if you can at all help it.

 

Do whatever it takes to figure out what ports are needed and use bridge mode.

So having my Network type set to "Custom: br0" is host mode, and that's bad?

I have another (vital) container that is using port 8080, and I don't think there's any way around that.

What would be the solution?

 

EDIT: Sorry for the 2 msgs, it wouldn't let me edit in a 2nd quote.

Edited March 5 by Elmojo
apology.

[JonathanM]

1 minute ago, Elmojo said:

So having my Network type set to "Custom: br0" is host mode, and that's bad?

If you have a different IP assigned to the container that's not host mode. A unique IP can open any and all ports without conflict.

 

Host mode uses the Unraid main IP and can conflict with Unraid's native services. Bridge mode is more like giving the containers their own router and setting up port forwarding where needed.

[PeteAsking]

Most of the issues here are people having no clue what networking decisions they are making and how networking works which is unrelated to this docker. 

[Elmojo]

49 minutes ago, JonathanM said:

If you have a different IP assigned to the container that's not host mode. A unique IP can open any and all ports without conflict.

 

Host mode uses the Unraid main IP and can conflict with Unraid's native services

Okay, now I'm actually more confused. lol

At least my setup is working, so I'm gonna leave it alone...

25 minutes ago, PeteAsking said:

Most of the issues here are people having no clue what networking decisions they are making and how networking works...

Yep, that would be me!

[NeuroDoc]

2 hours ago, PeteAsking said:

Most of the issues here are people having no clue what networking decisions they are making and how networking works which is unrelated to this docker. 

Could be me, but I'm always learning thanks to the great support on forums like this one!

 

...Now, if you need help understanding how to run international clinical trials in rare diseases, I'm your guy!

[bmartino1]

On 3/5/2024 at 11:23 AM, Elmojo said:

So having my Network type set to "Custom: br0" is host mode, and that's bad?

I have another (vital) container that is using port 8080, and I don't think there's any way around that.

What would be the solution?

 

EDIT: Sorry for the 2 msgs, it wouldn't let me edit in a 2nd quote.

Let's run through some basic networking and docker settings.
 

What are your unraids network settings?

I'm more interested in the setting for bridging and bonding...

 

 

Then what are your docker settings?

 

More interested in Docker custom network type: and host access.

 

This is my configuration for a proper custom bridge IP settings for macvlan network access.

 

Depending on how you want to use the docker network and how you want to access dockers from unraid we will need this information to know what type of network setup you are using.

read about docker network types here: https://docs.docker.com/network/ and https://docs.docker.com/network/drivers/

Custom br0 is a type of bridge driver...

 

 

By default, Unraid has bridging enabled and wants docker network to use ipvlan.

This changes how docker setups and communicates network traffic to the docker network.

 

If we run a docker image and set the network to host. The ip to access that docker if it has a webUI will be the same ip as the unraid system. Any ports the docker has will replace unraids ports and is more for items like clam av to scan itself or databases you want on the LAN to be accessible via your local network.

 

Bridge network mode uses the Unraids Default 172.x.x.x dhcp server and creates a switch port bridge to communicate via sharing your unraids network adapter. your docker is asigned a 172.x.x.x address in bridge mode and how it communicates depends on unraid docker setting, ie macvlan or ipvlan. But more to do with host network access.

 

Custom br0 means to bridge/bond the eth0 of your unraid and have it share a LAN ip address from your router. In my case my router dhcp is 192.168.1.x meaning the docker will get a 192.168.1.x unless I tell it otherwise. macvlan will make a random mac address to facilitate this, and some firewall/routers will see that your unriads ip is being duplicated or assigned to multiple devices sharing the same ip. (This is why the switch to ipvlan to fix layer 3 networking issues, buts only a problem with vlan network traffic controls...

 

If I run a macvlan network I can add an extra option and set a custom mac address to help router traffic. (this was required in older settings before "private mac" and phones were connecting to the internet to guarantee a network path access.

 

macvlan extra parm in docker : --mac-address 02:42:C0:A8:01:XX ->hexadecimal (0123456789abcdefgh) replace xx with a combination of a hexadecimal group...

 

Then the router can static assign.

 

Unfi has a host adopt feature that looks for a dns name unifi on the network you want the unfi network application to have a dns name set to unifi. So I also run macvlan to add hostname: extra parameters:

 

--mac-address 02:42:C0:A8:01:0A --hostname unifi

 

^This now give me the ability to statically assign an ip address. Via dhcp on custom br0 network which is the LAN DHCP IP from my router...

 

Unraid to use custom br0 it is recommended to static client asign an ip address. So custom br0 will be 192.168.1.x

 

this mean I can access the unif application at 192.168.1.x

 

Basic network that is dependent on what you run and how you run it. We need more info for troubleshooting network like these.

 

Edited March 7 by bmartino1

[Elmojo]

35 minutes ago, bmartino1 said:

Let's run through some basic networking and docker settings.
 

What are your unraids network settings?

I'm more interested in the setting for bridging and bonding...

Mine looks like this...

 

I feel like your post contains excellent, and extremely useful information, I'm just too ignorant to make use of it. lol

[bmartino1]

On 3/7/2024 at 5:36 PM, Elmojo said:

Mine looks like this...

...pictures...

I feel like your post contains excellent, and extremely useful information, I'm just too ignorant to make use of it. lol

No worries. It look like you are running multiple nics via a blade server. I recommend using a macvlan drive in your case. I would always recommend to use macvlan over ipvlan unless you require vlan network separations.

 

Per maclvan setup release notes: https://docs.unraid.net/unraid-os/release-notes/6.12.4/

 

I also recommend installing the plugin Tips and tweaks:

 

Goes over information about the settings.

 

Your system atm is using bridge mode and ipvlan for docker access. This should be fine. Surprised to see more than 5 nics on 1 box. 

 

However, bonding is similar to bridging in this case. We get a vhost@interface to act as a gateway between network interfaces. I would recommend you using bonding over bridging to maintain your network.
 

To do this, you will need to have some downtime to turn off some services of that box to edit these settings.

unriad webUI > settings> VM > disable

unriad webUI > turn off auto start of all docker (as there may be some template edits to fix networking...)

unriad webUI > settings> Docker > disable

unriad webUI > settings> NetworkSettings

 

I'd recommend turning off bridging and turning on bonding.

bond eth0 and eth1 as your past configuration for bridging the 2 interfaces.

bonding will also open up more setting for you for fail over , backup if one goes down etc...

 

Then we need to check and fix some docker settings.

unriad webUI > settings> Docker > advance switch

 

its fine to continue to use ipvlan docker driver. you should enable "Preserve user defined networks" and Host access to custom networks.

click apply with docker still off to make changes.

 

-?There may be a reboot required.

Then enable VM and Docker.

Confirm dockers by edit the docker template and fix network if needed. Br0 should be bond0 now.

[Elmojo]

7 hours ago, bmartino1 said:

It look like you are running multiple nics via a blade server

Yes and no.  Yes, multiple NICs.  No, not a blade server.  It's a PowerEdge T630.  It came to me that way.  I've actually removed 2 NICs to free up slots for other things.  No idea why they had/need so many.  I'm currently only using 1 port on a single NIC, but I'd love to be able to "gang" multiple connections to increase overall speed, kinda like we used to do with the old shotgun modems.  Is that what bonding does, or is it just for failover/redundancy?

 

7 hours ago, bmartino1 said:

Per maclvan setup release notes:

So I read this before upgrading, and understood about 10% of it. lol
What's a call trace?  I assume it's bad, but there's no explanation.  I've never seen anything like it as an error or in my logs, so I assume I don't have that issue?  You say you recommend using macvlan, but later you say that using ipvlan is fine.  I'm confused.  Can you give an example of what difference it would make either way, and what sort of consequences it would have leaving things as they are?  I'm firmly in the "if it ain't broke..." camp.    However, I'm also a believer in tweaking things if they can be improved.  Yes, I hear the irony... lol

7 hours ago, bmartino1 said:

However, bonding is similar to bridging in this case....

I'm sorry to report that I comprehended essentially nothing below this line.

I can certainly follow the steps outlined, but... why?  What would be the benefit?

I'm not doubting your skill or knowledge at all, I'm just wondering, what shortcoming is this addressing, or what positive feature would I be enabling by monkeying with settings that (as best as my ignorant self can tell) are working fairly okay at the moment?

ELI5, por favor? lol

 

EDIT:  So I did install the Tips and tweaks plugin, thanks!   Are there specific changes I should be making, using that plugin?

Edited March 10 by Elmojo
plugin Q

[bmartino1]

 

Your docker settings are trying to use a macvlan setup per your settings. But there is a known issue with macvlan and bridging on. Yes a call trace is bad, it's a system crash / freeze. They can be recovered and hidden with in dmesg log. its easier to use FCP plugin to scan to see if you have this issue.

 

If you want to use docker macvlan you need bridging off per release notes. Otherwise, edit your docker settings to use ipvlan. I apologize if that was not clear. I don't know how to be any clearer.

 

My understanding here is you have a network issue with this docker, and these are the next step to fix that network issue, which is why I gave the step to fix and set up the macvlan network... per the information you gave me.

 

Yes, bonding is what you could do to "shotgun" your connection. bonding is more for fail over. Read the tips and tricks for each bonding option type in the picture I sent. Read to better understand your network with bridging vs bonding. 

 

A bit abstract as unriad is linux but covers the gist:

http://www.linux-admins.net/2010/09/network-card-bonding-on-centos.html

 

https://www.cyberciti.biz/faq/ubuntu-linux-bridging-and-bonding-setup/

 

https://access.redhat.com/documentation/en-us/red_hat_virtualization/4.0/html/technical_reference/bridge_and_bond_configuration

 

Youtube: 

 

 

 

 

In terms of Plugins. Tips and tweaks is a great one for general use and knowledge of a setting. It adds informative information to the setting in unraid.

 

 I would also recommend these plugins:

if running latest 6.12.8
the Docker.patch2

 

CA Auto Update Applications > used to update dockers and plugin on schedule
Docker Compose > add compose to unraids docker system
Dynamix File manager > FCP tool for latter if needed
Dynamix System information > Linux system info dump to get hardware details
Enhanced log viewer > adds color and helps filter log
Fix Common Problems > Squid Common scan and general fixes
Tips and Tweaks > General Setting information
User Scripts > pearl/cron scripting when necessary for manually overrides and run sh scripts on schedule.

 

Edited March 10 by bmartino1