Profile cover photo
Profile photo
Bastian Blank
406 followers -
Debianer mit einem Hang zum Hardwareschrauben
Debianer mit einem Hang zum Hardwareschrauben

406 followers
About
Communities and Collections
View all
Posts

Post has shared content
Oh
"Removed" a photo series by Eric Pickersgill with digitally physically removed smartphones from pictures. Depressing. But also beautiful. http://www.removed.social/593ic8xxyj99ehv4sewgueg2ndwzod
REMOVED
REMOVED
removed.social
Add a comment...

Post has attachment
Da will man in den Bahnhof und dann steht da was im Weg. Also doch auf die andere Seite.
Photo
Add a comment...

Post has attachment
Irgendwie ist DB Vegetationskontrolle mal wieder nachl├Ąssig.
Photo
Add a comment...

Post has attachment
Formsignale sind schon ein seltener Anblick geworden.
Photo
Add a comment...

Post has attachment
#ingresspride everywhere
Photo

Hier gibt es eine B├╝ 4. Sehr oft sieht man die ja nicht mehr. (B├╝ 4 wird auch Pfeiftafel genannt.)
Ach ja. Die W├Ąscheleine fehlt auch.
Add a comment...

Did a few mini steps with Kubernetes. Just with GitLab and their Auto DevOps stuff. It does not work out of the box with current versions, because they did not manage to support clusters with authorisation (RBAC) enabled. But after hacking some permissions it actually works.

But what about using Kubernetes to do what I sometimes do as my day work? Aka moving a lot of mails via Postfix, Amavis and some other components.

Lets start with something stateless: Amavis.

It needs a lot of config that may change rather often. It seems the consent how to get configuration to the application is by using a config map. Changes to such a map are even applied to the files within running container. But there seems to be no way to notify the application of such a change.

Okay, Amavis needs a complete restart anyway to see config files changes. So a config map change could just trigger a re-deploy of the pods. Well, it can't and the helm FAQ even provides you with a trick to change the deployment every time the config map is changed. So this may work.

Then we have something stateful: Postfix. That's where it gets interesting.

Okay, we have config files again. Given that Postfix does not really like regularly restarts, doing reloads and map rebuilds would be a good idea. Given that there are no notifications, this needs to be done by hand. But maybe I could also use the restart pattern and just move all maps to some network type.

Also Postfix can only log via syslog, so it must have some syslog server, providing /dev/log. Can I use shared volumes to share such sockets between different containers of the same pod?

Also, do I want to look at helm for managing this?
Add a comment...

Jetzt ganz neu: DNS ohne Glue-Records:

gelsen.net. 172800 IN NS pns.gelsen-net.de.
gelsen.net. 172800 IN NS sns.gelsen-net.de.

gelsen-net.de. 86400 IN NS pns.gelsen.net.
gelsen-net.de. 86400 IN NS sns.gelsen.net.
Add a comment...

Post has attachment
Fahradweg, Symbolbild.
Photo
Add a comment...

Post has attachment
Add a comment...
Wait while more posts are being loaded