TLS certificates don't persist

[curious-broccoli]

"Problem" with hosting own instance

As stated in the wiki caddy requests a new certificate because the caddy data isn't persisted through restarts. Is this behavior intended? Because it shouldn't be much work to make the data persist. I think the issue is that the target location of the bind mount at

codabench/docker-compose.yml

Line 14 in ee8cdbc

- ./certs/caddy:/etc/caddycerts

points to the wrong location and an environment variable should be set so that it works, as stated in the caddy container's instructions.

On my fork I also upgraded to Caddy 2 but the change to make it persist was to simply fix the bind mount location (curious-broccoli@d8114bf) and it seems to work.

[Didayolo]

Thank you for reporting this. I think we had issues with this in the past, especially when we were restarting the server multiple times we encountered issue obtaining a Let's Encrypt certificate.

I thought this was resolved. At least, I think we changed the server configuration to avoid this problem on our setups.

We'll have a look at it. Thank you for the proposed fix.

[ihsaan-ullah]

@curious-broccoli why have you replaced ./certs/caddy:/etc/caddycerts by ./certs:/data

In the caddy documentation they are using /etc/caddycerts

[curious-broccoli]

Uh, the commit I linked is for the official caddy 2 image that I use on the fork.

The /etc/caddycerts is fine to use, but that example requires setting the environment variable to work.

CADDYPATH=/etc/caddycerts

as stated in the part of the docs you linked. So I think you just gotta set that.
Sorry for any confusion.

[Didayolo]

@ihsaan-ullah In conclusion, do we merge or close #1360 ?