Package: systemd Version: 246-2 Updating systemd to 246-2 leaves systemd-logind.service from 245-* which breaks certain operations.
One such example is, (assuming user foo): loginctl enable-linger foo. Without restarting systemd-logind.service this command fails with the following error: Aug 06 17:08:06 aug061652-827018 systemd-user-runtime-dir[28557]: Failed to acquire number of inodes for runtime directory: Unknown interface org.freedesktop.login1.Manager or property RuntimeDirectoryInodesMax. Looking at the systemd.postinst script I can see that systemd-logind is not restarted on purpose, referencing upstream systemd bug https://github.com/systemd/systemd/issues/1163. This bug is closed now so I suspect the proper action is to simply restart systemd-logind.service. I've attached a crude patch with that change.
--- /var/lib/dpkg/info/systemd.postinst 2020-08-03 09:46:27.000000000 +0200 +++ systemd.postinst 2020-08-06 19:33:31.845505560 +0200 @@ -114,8 +114,7 @@ if [ -n "$2" ]; then _systemctl daemon-reexec || true - # don't restart logind; this can be done again once this gets implemented: - # https://github.com/systemd/systemd/issues/1163 + _systemctl try-restart systemd-logind.service || true _systemctl try-restart systemd-networkd.service || true _systemctl try-restart systemd-resolved.service || true _systemctl try-restart systemd-journald.service || true