[feature] View content of volumes inside isaiah container.
I deploy isaiah via docker. When I try to browser volume in shell I get the following error and cannot proceed.
I think if we have mapped the docker volumes from the host to the container of isaish (e.g. /var/lib/docker/volumes:/var/lib/docker/volumes), we should be allowed to view the content of the voluems.
I learnt that the isaiah image does not contain a shell.
If I build the image based on alpinelinux/docker-cli:latest and do not set DOCKER_RUNNING, I can view the volume.
Thanks a lot for opening the issue, but most importantly for finding a flow that works.
I am currently looking into it, and running tests just to make sure that everything will be alright when released.
So far, this is how it will work when fully released :
-
The project's Dockerfile will no longer use
Scratch, but ratheralpinelinux/docker-cli, as you suggested and tried. -
The
DOCKER_RUNNINGvariable will remain unchanged, as it freezes other on-host-only features. -
The attachment of a volume to
/var/lib/docker/volumeson the container will allow thevolume.browsecommand to bypass theDOCKER_RUNNINGcheck, to enable users to browse their volume via the browser if the volume is attached to Isaiah's container. -
The image should remain tiny in size thanks to the base image you've suggested.
If you see anything else, please feel free to let me know.
Thanks again!
Alright.
I was busy lately, but I finally managed to tweak the code, run tests, and ensure that everything works.
I have just published a new release, and you should now be able to browse your volumes even when Isaiah is deployed in a Docker container, provided that you have previously added a mount that is identical to the one you've provided in your first comment (and that's also indicated in the error popup if anyone needs it!).
Hoping that it works better now, and thanks again for opening that issue.
Edit : I finally settled on the Busybox image, but feel free to tell me if there was an advantage with using alpinelinux/docker-cli.
It does not work on the node where "Agent" is running. (Multi-node mode). The agent hangs somehow.
Thanks a lot for your feedback.
I had not tested the feature in multi-node mode, it's my bad, my apologies.
I have just fixed the code to accomodate multi-node mode, and it should work now (or at least, it works on my end).
You can pull the latest image.
Sorry again, and looking forward to reading from you!
Thank you!