I have certbot including automatic renewal in use in several installations with clients.
Now I have been reading here:
https://community.letsencrypt.org/t/2018-01-11-update-regarding-acme-tls-sni-and-shared-hosting-infrastructure/50188
here
https://community.letsencrypt.org/t/solution-client-with-the-currently-selected-authenticator-does-not-support-any-combination-of-challenges-that-will-satisfy-the-ca/49983
and here
https://github.com/certbot/certbot/issues/5405
as well here:
https://community.letsencrypt.org/t/important-what-you-need-to-know-about-tls-sni-validation-issues/50811
everyone seems to be saying something slightly different
and no clear explanation is given.
I've been reading in the standard description of the certbot installation
Renewal docs point here:
https://certbot.eff.org/docs/using.html#renewal
but: the old vulnerable tls-sni-01
method is still listed
I try to summarize:
In the guides to mitigate existing issues on existing servers:
they recommend stop and starting the server on renewal. But... this is not good. What if some config breaks, and the server gets stop started when I am asleep? The server would be unavailable. Or something worse. I am not an devops expert but having the server more or less randomly start and stop, doesn't seem like a good solution. Am I wrong about this?
As alternative, I see only the webroot plugin that does not use tls-sni-01
. https://certbot.eff.org/docs/using.html#webroot
that seems to be the only way to me, that seems reliable.
Am I missing something? are we basically told to use the webroot plugin?
as all others do use tls-sni-01, are not automated (you can do it manually but ehrm, I actually don't want to) or require you to not have a running server already (standalone).
That is for future servers. I guess existing domain renewals will continue to work with the old tls-sni-01
, that is what they seem to be saying.