#NAbox GUI is unreachable after restoring the VM (following a non successful upgrade to 3.4 version)

1 messages · Page 1 of 1 (latest)

celest cloud
#

Hello all,

Yesterday afternoon I tried to upgrade NAbox from 3.2 to 3.4.
The GUI told me that it was successful, however Grafana was still running version 8.5.17...
Harvest has been upgraded as well from 2023-05 to 2023-11 with no issue this time.
After that, we noticed that in our dashboard there were no more metrics collected for Temperature, power consumption,...

I decided to perform a " dc down " and " dc up -d " but the first step failed with the following error:

" Network docker-compose_default Error 0.0s
failed to remove network docker-compose_default: Error response from daemon: error while removing network: network docker-compose_default id 35f00ff35218405076f74567f2e56d173869e883c2d107173723aac664ab44e9 has active endpoints"

The second step failed as well (see screenshot attached for details):

I tried to restore NAbox VM from the backup realized before the upgrade, it completed successfully but now I am unable to reach NAbox GUI.
I had a look at images and saw that " NAbox-api " is all the time in " restarting " status (see screenshot attached for details):

Any idea ?
Thanks in advance.

image.png image.png
clever sorrel
#

Sorry to hear that. There was another discussion pointing to the same error that seems to indicate NAbox isn’t finding the images locally

#

When you do dc down, there is one caveat : nabox will automatically restart everything so dc up -d might not be successful and conflicting with one already running through nabox-api-watchdog daemon

#

I’m assuming you don’t have the output of dc logs nabox-api

#

The failure to remove network is probably explained because by the time dc down finished, watchdog restarted the container locking the network and preventing your own dc down to remove the network after removing all the container so that’s fine

#

I would like to see the output of docker images to try to understand why docker images are missing. There is a cleanup mechanism to prune unused images that might be the cause. If it has been run while the stack was down, it would indeed remove critical images.

#

If that’s the case, you can download the update package manually and I can help running the upgrade in ssh

#

There is indeed a race condition I believe. When nabox-api start the stack (after an actual dc down and if the upgrade hasn’t been properly finished) it’ll prune image for containers that weren’t created yet

celest cloud
#

Hello @clever sorrel ,

Thank you for your feedback.

Please let me know where I can find the required log files and I will provide you with everything that you need, as I am not familiar with Docker environment.
We can proceed as well with any SSH manual upgrade as long as it solves this issue 😉 .

By the way, maybe a stupid question: do you think that NAbox/Harvest is still collecting data during NAbox " outage " ?

Regards.

clever sorrel
#

Please send me an email to help@nabox.org and we’ll schedule some time.

celest cloud
#

OK sure, email sent.