it helps to minimize the need for configuration if the default port is
used instead of a custom one.
Reviewed-on: #737
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Co-authored-by: Earl Warren <contact@earl-warren.org>
Co-committed-by: Earl Warren <contact@earl-warren.org>
The page grew over the past year to have redundancies and
inconsistencies. It is organized to keep sections that are common to
all machines in one place:
* LXC container creation
* Forgejo runner creation
* DRBD
* vSwitch
The other sections are unified (Disk partitioning).
Per "request" of @0ko in forgejo/forgejo#3270 (comment)
TODOs
- postponed: how to add codespell (written in Python) to CI -- separate workflow? add to some other step? to some other tool/specification?
Reviewed-on: #536
Reviewed-by: Earl Warren <earl-warren@noreply.codeberg.org>
Co-authored-by: Yaroslav Halchenko <debian@onerussian.com>
Co-committed-by: Yaroslav Halchenko <debian@onerussian.com>
They get /forgejo/forgejo/archive/100448a007707c049a769cb71cd170fcb5b64a25.bundle
despite https://code.forgejo.org/robots.txt having
Disallow: /*/*/archive/
Disallow: /*/*/archive/*.bundle
* Release number is v7.0.0+1.22.0 and following semantic versioning
* Remove the workflow page which is no longer needed as:
* it is implicitly understood pull requests go to the default branch
* testing requirements are documented in the testing.md page
* Explain how to cut a release and set the tag in forgejo-integration
for proper numbering of the test releases that come out of the
development branch
Refs: forgejo/forgejo#2448
instead of keeping a backup within the container itself, trigger the
backup external to the container to make sure it reflects the state of
the container after Forgejo is shutdown in case it needs to be
retrieved.
* switch to using lxc-helpers alone instead of enough
the learning curve of
[enough](https://enough-community.readthedocs.io) is an unecessary
burden in this context. Manual installation with lxc-helpers and
documentation is enough when there only is a need for a handful
of manual intervention per year
* update the Forgejo runner installation instructions
* tested the docker + self-hosted label to work with a test workflow
* do not maintain and setup a self-signed certificate because it creates a
complication that is unecessary when behind a VPN
No information in the root filesystem is really worth saving as it
can be recreated from this documentation or even from scratch manually
if needs be. But it is cheap to keep that backup and it is worth the
effort if it can save a few hours of work when an accident happen.
The more sophisticated way to do it would be to setup DRBD with the
root file system. But this is not straightforward and to ensure it
won't get the admin in trouble when an accident happen is tricky. It
is the one time when every bit of complication can quickly become a
blocker or an unecessary burden.
next.forgejo.org has been lagging for a few weeks in upgrades. While
crystal and dachary are busy, I setup and documented an LXC container
on the machine to which we all have access. It can be switched back to
where it was by updating the DNS entry.
