This makes sense for Redis but I don't like that we're setting options with system-wide implications in the redis systemd unit.

This makes sense for Redis but I don't like that we're setting options with system-wide implications in the redis systemd unit.

What do you propose we do instead?

What do you propose we do instead?

I'm not sure what the correct approach is, but what's currently proposed in this PR is counter-intuitive.
At the very least the Redis module should contain a note stating that this is a side-effect of enabling the module though I think that this is also not enough.

The Redis module should contain a note stating that this is a side-effect of enabling the module.

I extended the documentation of redis.enable accordingly.

@peti at least for me, the redis module is entirely broken, noticed during #67770. PR at #67845.

@flokli,

at least for me, the redis module is entirely broken

Can you be a little more specific, please?

The service fails to startup, see the gitlab vm log from #67770 (review).

The service fails to startup, see the gitlab vm log from #67770 (review).

Maybe I am blind or something, but to me eyes that snippet contains zero information about Redis. All I see are log messages from Gitlab, which is not the same thing, no? Anyhow, this is probably not the right place to discuss this topic anyway. #67845 probably is.

I'd really have preferred to have seen the fixes added inside the PR.

Adding more commits disconnected to the PR makes things much less understandable, especially if pushed to master without a reference in the commit message.

I added the small cosmetic changes, and included the release notes changes into #67845.