aboutsummaryrefslogtreecommitdiffstats
path: root/docs
diff options
context:
space:
mode:
authorGravatar Johannes Christ <[email protected]>2024-07-02 20:58:16 +0200
committerGravatar Johannes Christ <[email protected]>2024-07-07 17:45:35 +0200
commit80a4dd9c511c27736c7af791267ff9e12dc59245 (patch)
treeb81e7da86c6061c6c216607bf8e71ea4de55dcc5 /docs
parentUpdate meeting doc to correct date (diff)
Finish meeting notes for today
Diffstat (limited to 'docs')
-rw-r--r--docs/content/docs/meeting_notes/2024-07-02.md98
1 files changed, 95 insertions, 3 deletions
diff --git a/docs/content/docs/meeting_notes/2024-07-02.md b/docs/content/docs/meeting_notes/2024-07-02.md
index 5246f16..d84e2c2 100644
--- a/docs/content/docs/meeting_notes/2024-07-02.md
+++ b/docs/content/docs/meeting_notes/2024-07-02.md
@@ -18,16 +18,49 @@ Useful links
-->
+## Attendees
+
+Joe and Johannes.
+
+Chris unfortunately died in a fatal train accident and could not attend the
+meeting. This incident will be rectified in the next release, "Lovering 2.0:
+Immortability".
+
+Bella is out on the streets again. We are waiting for approval from the Python
+Discord admins to run another fundraiser.
## Agenda
+- **Configuration of renovate** (Joe)
+
+ We are replacing dependabot with renovatebot. Johannes welcomes this decision.
+ Joe says we are looking for automatic deployment from Kubernetes to make sure
+ that any updates are automatically deployed.
+
+ **Conclusion**: Implemented.
+
- **Resizing Netcup servers** (Joe, Johannes)
+ We can probably get rid of turing, assess what else we want to deploy on
+ lovelace, and then ask for a resize.
+
+ **Conclusion**: Create issue to move things off turing, remove it from the
+ inventory, remove it from documentation, power it off, then have Joe ask for
+ server removal.
+
- **Updating the public statistics page** (Johannes)
Discussing and showcasing possible alternatives to the current infrastructure
powering https://stats.pythondiscord.com via the
- https://github.com/python-discord/public-stats repository.
+ https://github.com/python-discord/public-stats repository. Johannes presents
+ his current scripts that cuddle RRDTool into loading data out of metricity,
+ Joe says we will discuss with Chris what to do here.
+
+ The likely way going forward will be that *we will open an issue to set it
+ up*, the setup will contain an Ansible role to deploy the cronjob and the
+ script onto lovelace alongside with the `rrdtool` PostgreSQL user.
+
+ **Conclusion**: Johannes will create an issue and codify the setup in Ansible.
- **New blog powered by Hugo** (Johannes)
@@ -38,6 +71,14 @@ Useful links
contributors would be nice for this, although currently everything is still
local on my machine.
+ Joe mentions that we don't need to take anything particularly similar to the
+ current Ghost theme, just some vague resemblance would be nice. Most of the
+ recommended Hugo themes would probably work. Johannes will check it out
+ further.
+
+ **Conclusion**: Try the [hugo-casper-two
+ theme](https://github.com/eueung/hugo-casper-two) and report back.
+
- **Finger server** (Joe, Johannes)
Joe recently proposed [the deployment of a finger
@@ -47,13 +88,65 @@ Useful links
to consider whether members will update their files regularly - we may want to
incorporate functionality for this into e.g. King Arthur.
+ Joe says that we shouldn't put a lot of development effort into it, it would
+ be simply a novelty thing.
+
+ **Conclusion**: This is a nice cheap win for some fun which should just be a
+ simple Python file (via Twisted's Finger protocol support or whatever) that
+ connects to LDAP (see Keycloak authentication server) and outputs information.
+ We could possibly integrate this into King Arthur as well, so the querying
+ workflow could look like KA -> fingerd -> LDAP, or people could use finger
+ commands directly.
+
+- **Keycloak authentication server** (Joe)
+
+ Joe mentions that we are deploying a Keycloak server because for some members
+ authenticating via GitHub is cumbersome, for instance because their GitHub
+ account is connected to their employer's GitHub Enterprise installation. We
+ could hook up a finger server to the LDAP endpoint. Joe also mentions that we
+ might want to set up e-mail forwarding from pydis addresses to users via the
+ user database that will be stored in Keycloak.
+
+ Currently we only have a Keycloak installation that stores items in
+ PostgreSQL. This installation can federate to LDAP - we would simply have to
+ settle on some directory service backend. Joe suggests FreeIPA because he's
+ familar with it (including the Keycloak integration). The problem is that it
+ doesn't work on Debian. The alternative proposal, given that we're saving
+ ~50$/month on Linode, would be spinning up a Rocky VM with FreeIPA on it on
+ Linode (we already have the budget) or ask Netcup for another VM. Ultimately,
+ the system to run FreeIPA would be something CentOS-based. One aspect to
+ consider is networking security: in Linode we could use their private cloud
+ endpoint feature to securely expose the LDAP server to Keycloak and other
+ services in Kubernetes, if we were to run it in Netcup, we would need to use a
+ similar setup to what we currently have with PostgreSQL.
+
+ Any Python Discord user would be managed in LDAP, and Keycloak has the
+ necessary roles to write back into LDAP. Keeping the users in FreeIPA
+ up-to-date would be a somewhat manual procedure. Joe's plan was to pick up the
+ user's Discord username and use `[email protected]` as their name and do
+ account setup as part of the staff onboarding.
+
+ **Conclusion**: Will wait for Chris to discuss this further, but we simply
+ need to decide where we want to run the LDAP service.
+
- **Flux CD** (Joe)
Joe proposes deploying [flux](https://fluxcd.io/) as a way to improve the way
- we manage our CI/CD.
+ we manage our CI/CD. We want the cluster to be able to synchronize its state
+ with the git repository. There are some manifests in the repository currently
+ that are not in sync with the cluster version.
+
+ **Conclusion**: Approved, Joe will create an issue and do it.
- **Polonium** (Chris)
+ Question came up regarding why the bot does not write to the database
+ directly. Joe said it's not perfect to have the bot write to it directly - in
+ metricity it works but it's not perfect. Chris probably had good reason:
+ separation of intent.
+
+ **Conclusion**: Approved, write to R&D for financing.
+
- **Rethinking Bella: Suggested measures to gain autonomy** (Chris)
Chris will present our current plans to biologically re-think and improve
@@ -63,5 +156,4 @@ Useful links
leave, we will send him the result of this voting via NNCP.
-
<!-- vim: set textwidth=80 sw=2 ts=2: -->