PeteAsking

I have updated the docker to post a popup message for now when installing instructing users what to do on first install. As far as I can tell, so long as you follow the initial setup instructions everything works fine after that. So summary: 1) Take backup 7 days on current controller. 2) Stop old unraid controller and turn off autostart. 3) Set directory permissions (can be done after deploying image if preferred, see notes). 4) Go to apps and install new controller (unraid-controller-reborn) 5) Start container and set to autostart. Ensure it can write files and functions normally and you can change/update it etc as required. Restore backup, will require you to manually start it again after backup is restored. It seems pretty basic, just dont have an immediate fix for the "fix directory permissions" issue at this moment. It is a one time thing, however, not needed ever again.

Ok will take a look, also the container has updated with unraids repository so can be tested, albeit with the instructions to fix the directory.

Ok so for now I can see the issue but not figure out how best to resolve it. There are several ways but none are automatic. The easiest ways I found to change to this image are like this: 1) Take backup 7 days on current controller. 2) Stop old unraid controller and turn off autostart. 3) Go to apps and install new controller (unraid-controller-reborn) 4) Once it is installed, stop the container (it cant write files to the directory unraid creates so does not work) 5) Either ssh into the box and use chmod like "chmod 777 /mnt/user/appdata/unifi-controller-reborn" OR if you dont like that go to Tools - New Permissions then choose "shares" and "appdata" and let unraid fix the permissions (unclear why this is not done by unraid when creating a new appdata directory, also unclear if this tools-new permissions thing is safe/supported?). 6) Start container and set to autostart. Now it can write files and functions normally and you can change/update it etc as required. Thoughts on this? Kind of unsure why unraid behaves like it does. Not really an expert obviously on unraid and its docker implementation -P Edit: also just waiting for unraid to update my xml settings so I have to wait for that to complete before anyone can test it. Edit2: Wow so apparently unraid has a web terminal so you can just open the web terminal and copy and paste the following 1 line at a time before installing the app: mkdir /mnt/user/appdata/unifi-controller-reborn chmod 777 /mnt/user/appdata/unifi-controller-reborn chown nobody /mnt/user/appdata/unifi-controller-reborn chgrp users /mnt/user/appdata/unifi-controller-reborn Instead of these 4 commands you can also try: install -d -m 777 -o 99 -g 100 /mnt/user/appdata/unifi-controller-reborn That also seems to work. Unsure how to make that happen automatically at this time.

I think you might have just helped me figure it out. I will go change something. You were totally wrong but I like the way you think.

Yes but I think that only works if the docker maintainer has actually coded to accept those variables which is not the case for this image and it tries to use 1000:1000 and is hard coded into the image.

So made some progress but still stuck on this last part because I dont understand unraids docker enough. Basically you can deploy the community app "unifi-controller-reborn" now and when you deploy it it will deploy but not start. To fix it you have to stop the container after, then ssh and go to /mnt/user/appdata/ and do a chmod 777 unifi-controller-reborn/. Now it can write to the directory and everything works and saves correctly and it works like a normal container. What I can work out is how to get unraid to initially set the permissions on the folder when creating the docker image so this part Im stuck on now. - P

Yes when its working to my standard of quality. Also if this is the direction people want to go. If everyone votes to go with 2 containers then that would be the direction instead. Have to please the community.

There is now an ALPHA community app that you can test if you would like to. Instructions - 1) Login to current linuxserver.io unifi controller and take a backup (7 day backup seems reasonable?). 2) Turn off current linuxserver.io unifi controller and disable autostart. 3) Download and install new community app "unifi-controller-reborn" and enable autostart if desired. 4) Restore settings from backup file after logging into the app, once it stops the controller you will need to start it again. Current issues: If you change any settings whatsoever on the docker container all settings are lost and you have to restore again from backup. Im working on it. You have your backup file and the old container is untouched so presumably between those 2 options you can go back if need be. Many thanks P

Just an update, still having real trouble communicating with Squid. I will try reach out to someone else. - P

Thank you for doing this and showing how to do it. This is exactly what we need. Multiple options the community can collectively decide on. So long as we dont get super attached to a particular solution and let everyone collectively decide which is a better option then we can go all in one direction or the other when the time is appropriate. Currently I have everything in place except a way to get a community app published for unraid as Im having trouble getting hold of squid who is the unraid person we have to speak to to get something published. However I imagine he is just busy which is why we havent heard from him. Either way Im not particularly worried about getting a good solution in place by January. Since we have 2 potential paths forward it seems likely that one will be fruitful by then and everyone can then start a new forum thread and support each other going forward. Unraid is very flexible and it seems capable in doing pretty much whatever you want.

I will try if no reply by tomorrow

I did like 4 posts ago.

Not sure how to contact squid from these forums I sent him a PM and it says he has logged in since but he didnt read my message. If anyone knows please let me know as I cant progress without his help

Still waiting for a reply from Squid on these forums so I guess he is away or something as he hasnt read my messages yet.

I have been speaking to the developer and he told me he has a github page here: https://github.com/11notes/docker-unifi and he will be keeping it up to date going forward so if we have issues we can open an issue there. One thing he told me is that uid and gid should be 1000. Im not 100% clear on this so I have asked @Squidbut he hasnt gotten back to me on this. I also asked if we could make a community app and test it works but will have to wait his reply. So far this isnt seeming to bad but will post any new information I get as it comes. - P

I dont mind, but I like to make sure I have everything in place before making promises. So you will have to wait a few days while I test and discuss with the developer of the actual image - because if it transpires he is not accommodating to us using the image then I would be in a position where I have made promises to you guys that I cant keep. So at this point I am more interested in checking what can be done technically and then ensuring after that this would be a workable long term solution. Then I can come back here and tell you what I have found out so you can be involved in deciding what direction is best overall. Seems like it affects a lot of people so probably best to ensure everyone had a say. Also it is certainly the case that a valid approach is the one that linuxserver.io has suggested - 2 containers which has mongodb split from the unifi container. They didnt give any detail why this is better for them so I only see one side of the story here. It might be that unraid is different to most users of the images since most users are using podman or docker compose while unraid is just simple pure docker. Hopefully @wgstarkscan get a second option up an running so then you guys can decide at least between 2 options. Either way I will continue working on a solution as well and let you know what happens but Im pretty much confident that since we have until January to decide that there is no rush and one way or another something will be created by then. It just doesn’t seem like something everyone has to panic over so lets see what people come up with and let everyone decide what they like best based on some actual testing. End of the day we all want something that doesnt break and things were working pretty well I thought up till linuxserver.io decided to depreciate the image so just want to get back to that situation where things can be deployed and not have to worry much. Hope this makes sense. Just want to reiterate Im not promising anything until I feel confident things are up to my standard so I dont let anyone down on anything.

Yes that is exactly the sort of issue that could occur and why I dont personally want to go in that direction.

Its not my image but yes its identical to what the old soon to be depreciated image is in pretty much every way which is why using it is so simple. Its just picking up where linuxserver.io left off.

Its more complex to setup and requires 2 docker containers running and communicating with each other. So generally as complexity increases there is more that can go wrong and lower reliability is expected. Also nobody has started looking at how to migrate to it but you could take a look and tell us if its super easy.

Its not a big deal @wayner I am working on an easy solution. As you are worried let me put your fears to rest and let me tell you where I have gotten up to. We will need to involve the support of a few people on this unraid forum for a final solution but I have a 7 step migration plan that will keep you working and hopefully we can flesh out a template for the "community apps" so that it can get its own official thread. I have already started testing and moved to a different container and it seems to work fine. At any rate here is a work in progress if anyone else would like to test it as an initial test (for a final solution a backup and restore similar to below will be required again and it is very easy so if you can test please do so). A user I engaged with on reddit was in the same predicament and created his own container to resolve the problem. His is also using ubuntu as a base (same as linuxserver.io image) so is pretty much identical. DO NOT UNDER ANY CIRCUMSTANCES SIMPLY EDIT YOUR CURRENT RUNNING UNIFI CONTROLLER AND JUST REPLACE THE IMAGE REPOSITORY. His container is here: https://hub.docker.com/r/11notes/unifi STEPS TO CHANGE AS AN ALPHA TEST (Should not be dangerous to do): Step 1: Login to linuxserver.io running unifi controller Step 2: Take a backup in the interface (System - backups - Download - last 7 days) Step 3: Shutdown linuxserver.io controller (leave it off and available in case you need to go back). Disable autostart. Step 4: Create new container from template (Add container - unifi controller) Step 5: Change 3 values (name change "unifi-controller" to "unifi-controllerNEW" ; repository change "lscr.io/linuxserver/unifi-controller:version-7.5.187" to "11notes/unifi:7.5.187" ; change under more settings ‘appdata path to have ‘NEW’ on the end eg: /mnt/user/appdata/unifi-controllerNEW ; leave all other settings alone. (Note if you fail to change these 3 settings you will wipe your old container.) Step 6: Hit apply (It is created from a template exactly the same as the old settings, only thing left is to restore from backup just created). Step 7: Restore from backup after logging in (eg: https://192.168.2.4:8443) and set docker to autostart on unraid. You can now login and site is restored. As always, please take a backup before just following some randos instructions on the internet. Im trying my best but not infallible, if you manage to find some way to blow up your house following the above steps, there isnt much I can do so just be logical, check you understand whats going on and so on. Benefits of this method: 1 docker container is maintained going forward, very easy to migrate to, pretty much no customization beyond giving it a new name and pointing it to a new image location since it uses all the same settings as the old template Any feedback on this idea appreciated. Currently settings are not preserved on reboots of the image so this is a test only before you revert to your old container (I think, not sure still testing this). Kind regards P

I think you have to just setup 2 new containers (one for the unifi and one separate mongodb container). What I would do is change the default management port on the existing one and then try get the new 2 containers working together and when they are working then doing a backup from the old on and shut it down, and then import to the new one. Sounds like a lot of boring effort.

If you backup your appdata I believe you should have no problem upgrading directly to 7.5.187 but will be able to roll back if you need to. In this case you can tell us if the new version is acceptable to you and I will update what is old stable and bring the versions tagged in my previous post along one extra version so everyone can update to the next version applicable to their deployment. I have just deployed 7.5.187 and have had no issues. The upgrade was very fast for me also, a couple minutes at most. I note a couple of people on the forums at unifi had to restart a device that dropped offline manually in order to have it brought online to the controller but nothing more serious than this. Let me know if it works for you and I will update. This version seems like it will be the one to stick on for a while. Kind regards Pete

A new version UniFi Network Application 7.5.187 is available. I intend to update tomorrow, it seems like an alright version. We will find out tomorrow sometime if that is true.

This question has too little information to help you. Please add more relevant information.

I believe you can export site function from your config so you would do this process so you can readopt the AP's. https://lazyadmin.nl/home-network/migrate-unifi-controller/ also: https://www.clevendo.de/hilfe/support/setting-up-and-managing-your-unifi-controller-in-the-cloud/migration-en/moving-the-unifi-controller-to-the-new-server/ I havent had to do it but this process seems logical to me. When deploying via unraid the tag you use will be like this format: lscr.io/linuxserver/unifi-controller:version-7.4.162 Please backup your docker images with : Appdata Backup plugin. Kind regards, Pete