[6.11.5] All* Docker containers unable to configure following setting change

[Outlaws]

Hello Unraid wizards,

 

I come to you again in my time of need 🙏

 

I will spare you my full recollection of the morning, but please ask for any other details that would be helpful. I'm happy to troubleshoot. After a file share server, I primarily use Unraid for a number of Docker containers. I was looking in to improved monitoring for my home network, and had recently been playing with SNMP tools at work, so I was looking to enable something similar here. I found and installed some containers which looked relevant, and got those setup. When I was doing that however, I've somehow managed to break the ability to edit any other docker images I had previously? Here is a list of things leading up to the issue, and troubleshooting steps I've tried so far.

 

Scenario:

1. Searched for & found relevant docker apps for system/network monitoring

2. Installed MariaDB & LibreNMS --- these are the only two docker containers that I am still able to edit. They are also the only 2 working on the 'eth0' interface, PiHole and Unifi controller are currently no longer starting

3. Oh cool, this works. Began adding switches and other devices to LibreNMS, wanted to add Unraid as a server to be monitored...

4. Found that the Unraid host was unable to communicate with the Docker containers, and this was by design.

5. Stopped all running docker containers

6. Under Settings -> Docker, changed 'Host access to custom networks' to 'Enabled'

7. Went back to Docker tab, and uh oh. Some won't start, but all except the two newest are now unable to be configured.

 

Troubleshooting:

1. Step 1 in a bad change, undo the change. After observing the error trying to start PiHole, I disabled the containers and reverted to Host access to custom networks: Disabled.

2. Restarted the Docker service, but no change. Found that the containers are able to start except for PiHole and Unifi, but none except the two newest (MariaDB and LibreNMS) are able to be configured via the Unraid GUI.

3. Disabled containers, stopped array, reloaded server. Logged back in, but found that the issue persists.

 

Issue:

1. No longer able to configure/edit any Docker containers, except 2 newest

2. Unable to launch 2x existing Docker containers

 

Thankfully, I also have a CommunityAppData backup from this morning before everything, if restoring that would be helpful. As for why the two specific containers will not start at all, it seems to be mad about something with the network?  When I look at the Unraid console, 'docker network instpect eth0' shows me definitions for the interface, mariadb, librenms.... but not the .7 or .8 values I expect for Pihole and Unifi. I can also generate:

 

root@Unii:~# docker -D start unifi-controller
Error response from daemon: network d824fbdfadf6027a607225e63a07f64cd92447e9197bb92efd287652167a91a5 not found
Error: failed to start containers: unifi-controller

 

But the same command is unable to recognize anything for the Pihole container, which should be pihole-dot-doh or pihole-dot-doh-2. I can get the Pihole container to appear in other outputs though, such as '~# docker image inspect devzwf/pihole-dot-doh' and '~# docker image history devzwf/pihole-dot-doh'. Before I continue down the rabbit hole myself:

 

I've included some relevant diagnostic files and images of the issue, can share anything else that is needed. Please let me know what the best next steps would be?

 

maybe something like 'recreate docker.img' from here?
https://docs.unraid.net/unraid-os/manual/troubleshooting/#How_do_I_recreate_docker.img.3F

update:

1. moved this to the correct subforum

 

2. Deleting and recreating docker.img kinda works, but I'd love to avoid it. I can re-install exactly one container. I don't remember all the docker container configs, can't easily view them now, and trying to 'restore docker apps' like mentioned in the link below doesn't work - it only shows the past few weeks of installed apps, not everything. So..... can anyone help me avoid fully rebuilding my docker setup

Unraid_Diag_Files.7z

[Parmstig]

I think the unify-controller when setup gets identified with the specific network adapter, if you make changes to network settings in the container then that network adapter is not present anymore, maybe try with Bridged network so it can access the same network as unraid. 

[Outlaws]

Thanks for the response, Parm. I ended up nuking the docker.img and re-installing all the containers. Thankfully the CommunityAppdata backup helped retain all of the /appdata information.

[Nasha]

For all future reference - 6.12.4 is said to negate any nerfing that is initially undergone....

 

I'm not sure what actually nerfed the existence of a particular set of repeatable circumstances for very few people but all sharing common ties to widely used products independently (being Unraid and Docker and PiHole). For me, I thought that I'd launched the nerf. I now know that was pure coincidence

 

Pretty sure that given Unraid comment vaguely advising on upgrading versions in imperfect English, hintd to something that changed within Unraid, and tight to the dev-core (because of limited end-user impairment, and the correlations I know to have existed between public reports of issues around the time my problem occurred) - but I've not actually tried to "find the reasons for the cause", because I trashed most any semblance of change tracking the moment I manually manipulated related data. 

 

My upgrade was the installation of what I believe to be the root of the 6.12 tree and back to early 11s, up to 6.12.3 level was my last update&reboot performance. 

My next available reboot just happened to incur some possible changes to the Docker folder storage location on physical disk, so I chased it as a problem I had created, and tried to patch it my copying files between disks etc. to no avail, and I'd resigned to life without Docker on Unraid, until the mrs tried to use Plex and all hell broke lose. So I tried "starting again", gut, disable, enable from scratch, load items based upon predictable results, and struck luck with joining data dots with some others, that created the "perfect storm" scenario, which I may have completely avoided if I'd waited a minor revision longer also... 

 

So yeah just wanted to say, I felt it too, it sucked to have been the unlucky few, to face rather dire failures, but it appears that 6mths later it seems to have passed on by like a gentle breeze. Not the catastrophic nightmare that was having Docker disabled from use for months irreconcilablely...

Thankfully, nothing that cannot be replaced was impacted, but it did place me at a critical DR scenario of single point failures for an extended period of time.

 

But an important reminder that unraid shouldn't be relied upon to function beyond servicing your data storage requirements and interists, certainly not concerning the excess of separation from common interest 3rd party plugins to a 3rd party application that becomes a nexus for 3rd party integrations in general to becomr a convenient nexus to apply integrations that this issue created/brought to attention... Including the reminder that removing single disk failures from your life is by no means = to safe and comprehensive means of data storage full stop. Learning the hard truths about multiple disk failures, and how easily they can occur is extremely painful, trust me!