postgresql: Make systemd support configurable

[nh2]

Motivation for this change

See #61580.

On by default, as was before. No evaluations should change.

The systemd dependency means that all libraries or applications using only libpq
will also pull systemd into their closures.

Further, no application or library using libpq can be built against musl, as
systemd relies on glibc-only features and thus does not build with musl.

With it being configurable, packages that need only the library can at
least turn it off to reduce their closure size.

Things done

• Tested using sandboxing (nix.useSandbox on NixOS, or option sandbox in nix.conf on non-NixOS)
• Built on platform(s)
  • NixOS
  • macOS
  • other Linux distributions
• Tested via one or more NixOS test(s) if existing and applicable for the change (look inside nixos/tests)
• Tested compilation of all pkgs that depend on this change using nix-shell -p nix-review --run "nix-review wip"
• Tested execution of all binary files (usually in ./result/bin/)
• Determined the impact on package closure size (by running nix path-info -S before and after)
• Assured whether relevant documentation is up to date
• Fits CONTRIBUTING.md.

---

[thoughtpolice]

Based on the discussion in #61580, I'd restructure this all to look something like this:

{ version
, useSystemd ? true
...
}:

let
  # systemd doesn't build with musl, and can't work on darwin
  # see #61580 for information & discussion about musl support
  systemdCompatibleEnv = !stdenv.isDarwin && !stdenv.hostPlatform.isMusl;

  # only postgresql 9.6 and later support systemd notifications
  systemdServerSupport = stdenv.lib.versionAtLeast "9.6" version;

  # enable if asked, iff the prior conditions are met
  shouldEnableSystemd = useSystemd && systemdCompatibleEnv && systemdServerSupport;
in
  # ...

  buildInputs = ... lib.optional shouldEnableSystemd [ systemd ] ...

[thoughtpolice]

(See prior notes)

[thoughtpolice]

And, actually: in this case, of adopting my suggested change from the last comment -- I probably would just not even include useSystemd ? false as a parameter, since there is, AFAICS, little reason to avoid it otherwise.

I wouldn't block this change if it was left in -- it's just that I don't think it's very useful at all, given the proposed change, and if we add it, we can't easily remove it later since someone might use it (even for odd or strange reasons), effectively locking us into supporting it.

[danbst]

This shouldn't harm IMO.

[danbst]

maybe also add an assert that systemd can't be null when enableSystemd = true. Though this is minor

[fpletz]

This breaks systemd support for postgresql on master resulting in the systemd service being killed due to timeout because postgresql won't notify systemd, see this journal log:

1563849511.732376 trolovo systemd[1]: Starting PostgreSQL Server...
1563849511.786683 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  database system was shut down at 2019-07-23 02:34:59 GMT
1563849511.791724 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  MultiXact member wraparound protections are now enabled
1563849511.792795 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  database system is ready to accept connections
1563849511.793021 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  autovacuum launcher started
1563849631.761013 trolovo systemd[1]: postgresql.service: Start operation timed out. Terminating.
1563849631.763162 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  received fast shutdown request
1563849631.763162 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  aborting any active transactions
1563849631.763837 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  autovacuum launcher shutting down
1563849631.763948 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  shutting down
1563849631.809723 trolovo g8fqi5ssqms72x760xg3njcgncp1b27w-unit-script-postgresql-start[19562]: LOG:  database system is shut down
1563849631.810316 trolovo systemd[1]: postgresql.service: Failed with result 'timeout'.
1563849631.810877 trolovo systemd[1]: Failed to start PostgreSQL Server.

[fpletz]

Fixed on master (290cfc7). versionAtLeast was called with arguments in the wrong order.

Note that even the NixOS tests for postgresql fail with the original change that was merged. Please run NixOS tests as part of PR reviews.

[danbst]

sorry, will do