error: Failed to retrieve container image info: Error: No such image

[bertdeelman]

Hi,

I use Watchttower at a Synology NAS. I use the default configuration. After starting the container I receive two (errors/warming)

• Failed to retrieve container image info: Error: No such image: sha256:.....
• watchtower cron: panic running job: runtime error: invalid memory address or nil pointer dereference

Any idea what to do?

Regards,
Bert

[piksel]

Are you using docker compose by any chance? Not docker-compose, but the "v2" integrated command. It doesn't store the image name in a retrievable manner and does not work with watchtower unfortunately.

[bertdeelman]

Hi,

I don't quite understand what you mean.

I tried both. So from the command line:
$ docker run -d \ --name watchtower \ -v /var/run/docker.sock:/var/run/docker.sock \ containrrr/watchtower

And a with docker-compose.yaml file, with the docker-compose command. Same notifications.

[piksel]

I mean for the other containers on your system.
If you run docker ps, does all the containers have a name in the IMAGE column?
If it's instead a hash, it causes the above error in watchtower (and is caused by docker compose as far as we know).

#1027 addresses this, but only skips containers like that, it still cannot check/update them.

[bertdeelman]

ok, they have a name not a hash like the error message. I've got some strange image names like: 7c6d73bed103 or c060f9cb7fd5, but most of them are normal like bastilimbach/docker-magicmirror:latest etc.

[piksel]

Yeah, those are hashes 😁 (its the "short" format).

containrrr/watchtower:latest-dev contains the patch that should allow watchtower to skip those instead of crashing, but it's pre-release so it could be unstable. If you have any idea how you ended up with those containers it could be helpful.

[bertdeelman]

Hi,

I updated all images with a hash in their names by hand and use the latest-dev version of Watchtower. And for now it has successfully upgrade containers.

Thanks!

Regards,
Bert

[billouetaudrey]

@piksel I have the same issue " Error: failed to retrieve container ... " and he found the fix so I ask how he update all images ;) It's not off topic

[piksel]

@billouetaudrey Fair enough. I think he simply means removing the containers and re-creating them again using docker run or docker-compose up. You cannot change the image of a created container.

[billouetaudrey]

Ok thanks for reply :)
Good luck :)

[LostOnTheLine]

I just ran into this issue, but the comments here helped me figure out how to fix it.

My issue involved a container having a "hash" name, but it was the result, I believe, of an error, or perhaps the server froze or restarted during a watchtower update.
I went into Portainer & found the container, then stopped it's stack & restarted the stack. After I did that I stopped getting the error as it now used the updated image for that container instead of the old one without a name.

The Docker PS command was helpful, but if you use Portainer going to the Containers List would be more helpful, especially if you have it set to update containers that are not currently running

[LostOnTheLine]

Just verified that just deleting any containers that have a hash name fixes this as well as I just saw it for a container that was currently stopped & just deleting it worked.