sregistry under https about sregistry

[vsoch]

Is it possibile to push an image using singularity push without https?

Yes, this is the default, and it should work out of the box, with the caveat you usually need a special compiled version of the client.

When I made sregistry it meant compiling singularity with an adjustment to a file, which is described here https://singularityhub.github.io/sregistry/docs/client#singularity-push. That code, however, I think is old, and my guess is that now it's controlled by how you login to the remote: https://github.com/sylabs/singularity/blob/f1477058a25cddedb0ca5b2377b6f9ef827865a7/internal/pkg/remote/credential/login_handler.go. It's 1:30 am here so I can't look into this for you, but if you don't figure it out I'd ping the Singularity slack (Sylabs) and ask about how to get http for push.

For pull, they now have a no https --nohttps flag . With push I think it's still required, and you'll need to compile this special version.

Otherwise, how can I run sregistry using https?

This is what I'd consider a production deployment. We have ample docs here: https://singularityhub.github.io/sregistry/docs/install/https for both faux (development) and production https.

[imerelli]

Ok, thank you for the suggestions.

I'm trying to move sregistry under https. I was able to generate the certificates (that are now on the host server) and I moved the https/docker-compose.yml and https/nginx.conf files to the sregistry root directory (moving the other to http as suggested). I also modified the indicated section in the config.py as follows:

DOMAIN_NAME = "https://sregistry.bioinfotiget.it"
DOMAIN_NAME_HTTP = "http://sregistry.bioinfotiget.it"
DOMAIN_NAKED = DOMAIN_NAME_HTTP.replace("http://", "")

The docker compose run smoothly but the nginx server is not running properly,
because the site is down and also checking the port the server is not running:

nmap localhost
Starting Nmap 7.70 ( https://nmap.org ) at 2022-09-09 13:56 UTC
Nmap scan report for localhost (127.0.0.1)
Host is up (0.000012s latency).
Other addresses for localhost (not scanned): ::1
Not shown: 995 closed ports
PORT STATE SERVICE
22/tcp open ssh
25/tcp open smtp
111/tcp open rpcbind
631/tcp open ipp
8090/tcp open opsmessaging

Any hints on what the error can be? Is there any log I can check?

[vsoch]

Can you check the nginx logs? usually docker-compose logs or docker logs against the container can give us a hint.

[imerelli]

Looking at the logs it was clear that the file dhparam.pem was missing.
Thanks

[vsoch]

Sure! You reported https working in your other thread so let's close this issue.