This commit switches Hugo to a webhook-based building process, with
support for Github, Gitlab, and Gitea hooks (including local versions of
Gitea) initially. In addition, Hugo-based sites are now intended to be
served under a single volume, with ingress configuration pointing to
sub-paths into the volume.
Documentation for webhook setup and NGINX proxy configuration is still
underway, and will be filled in later.
This commit unifies the `UPSTREAM_HOST` and `UPSTREAM_PORT` environment
variables to a new `UPSTREAM_ENDPOINT` variable, making additional
customizations (such as a `proxy_pass` to a sub-path) possible.
Users in the `prosody_user` and `prosody_admin` groups will be granted
access to Prosody (as regular users and administrators, respectively),
making this a more flexible solution compared to IMAP.
New Gitea installations will now use LDAP authentication, typically
provided by the included `lldap` service, over SMTP authentication, as
this is is more flexible.
This commit extends our `coreos-home-server-update` script with support
for updating host directories with configuration collected across
multiple remote directories of the same name. This will, essentially,
allow for extending systemd services with custom configuration, as
sometimes required of base systemd service files.
This sets the stage for moving common authentication from IMAP/Dovecot
to LDAP, which allows for more control over user information, as well as
a basic form of RBAC.
No services are currently set up to support LDAP -- support will follow
soon after this commit.
This commit removes host configuration that is only useful for my own
personal use-cases, and leaves base and virtual hosts, which are more
commonly useful. It is intended that additional hosts are tracked as
either submodules or subtrees, as required by individual use-cases.
This commit adds a new service for Gitea, exposing HTTP and SSH ports by
default (SSH over 7920), and accepting authentication via the local SMTP
server. No users are otherwise created by default, and administration is
expected to happen either via CLI, or via a custom admin user.
Container volume backup logic has been moved to a (largely equivalent)
external script, allowing for future expansion of functionality. In
addition, a `rclone-pull@` service has been added and set up as a default
dependency for the `container-volume-restore@` service, allowing for
automatic set up of servers based on latest remote backups.
This commit updates Postfix to the latest version available for Debian
Bullseye, and switches the Docker entrypoint from an internal to a
publicly documented command, which is guaranteed to work in the future.
In addition to changes made when Git repositories update, Hugo is able
to watch for timed changes to content (say, for content that has a
publication date in the future) and build these automatically.
This commit switches from a plain `hugo build` to a `hugo watch` process
to facilitate these use-cases.
These were prepared as import-able dashboards, which prepares common
options as variables to be provided by the user, and thus is not
compatible with automated provisioning.
Users and groups used within Podman containers are usually assigned UID
and GID 10000. Files for these containers are sometimes served by Nginx,
and may be given permissions that restrict access to those outside the
group, but which are intended to be served nonetheless.
This commit adds the pre-defined `nginx` user to a `nginx-shared` group
with GID 10000, which will then allow access to these files as needed.
This enables the `web_directory` and `web_url` options for Spectrum,
which has media shared by legacy protocol buddies be re-hosted by
Spectrum, and shared by a dedicated volume.
This, in turn, is served by the `nginx-serve-volume@spectrum-media`
service, which is intended to be proxied under the same host used for
Prosody. Documentation for the integration will be added in future
commits.
This commit adds basic Grafana dashboards for Dovecot and Prosody, to be
automatically deployed alongside the relevant services (if Grafana
itself is enabled).
As the former does not do exactly what it says it does in documentation.
Also, we decrease the default scrape interval for Prometheus from 1m to
30s to improve granularity of data.
This adds basic metric definitions across a number of different events,
to be expanded upon as needed in the future. Several metrics are given
additional vectors with low cardinality.