Note to self: I should really stop bullying my Kubernetes cluster network (without proper preparation) ​:neocat_think_woozy:​

I'm getting very rare and intermittent timeouts/connection resets on the stuff I self host ​:neocat_woozy:​ only happens once on cold starts (when it happens)
Question is: is it caused by the recent addition of ECMP routing, by me having made tons of connections and disconnections in my network recently, or by random local network issues ​:neocat_googly_woozy_laptop:​

I acquired a new network switch, which is actively cooled ​:neocat_chew_wire:​ "nah it's gonna be fine" I said as I powered it on for the first time ​:neocat_mug:​
industrial grade fan bearing noise it wasn't, indeed, fine ​:neocat_shocked:​

Anyway, I acquired a new Noctua fan ​:neocat_3c:​

You're officially in the reservation queue for Steam Frame 1TB Kit.

Looks like Tomorrow is finally today ​:neocat_vr:​

Feels good to finally take the time to write a blog article about the thing that's been in the back of my mind since last December ​:cursed_cool:​

I'm very very happy with the fact that since I've migrated my homelab to K8s, I've spent a lot less time maintaining it and keeping up with it ​:neocat_comfy_mug:​ definitely didn't expect that increasing complexity could result in improved efficiency ​:neocat_comfy_happy:​

Turns out, when an upgrade note says "required" on a step, it is indeed required ​:neocat_boop_woozy:​

​:neocat_flop_sleep:​ Going to the train station on time to welcome someone
​:neocat_fingerguns:​ Going to the train station an hour early to take some pictures beforehand

Co-authored-by: the voices in my head

Whenever I build a Docker image using DinD on my k8s cluster, the node running the pod crashes and reboots ​:cat_confused:​

I wanted to monitor my Mikrotik routers with an SNMP exporter, but didn't really know where to start and didn't want to invest a lot of time into it, so I searched for ready-to-use solutions. ​:neocat_mug:​
Found some, deployed, was quite satisfied. ​:neocat_thumbsup:​
"Wait, this metric doesn't seem to work, lemme check that ​:neocat_think:​"
"But I don't have the SNMP exporter generator file... Fine, I'll rebuild one myself. ​:neocat_science:​"
"Uuh, this metric on the dashboard doesn't seem to use the right identifier... lemme fix that real quick ​:neocat_laugh_sweat:​"

"OK fine, I'll do it myself then! ​:neocat_scream:​"

This server is straight up cooking! ​:cursed_cool:​

And by cooking... I mean... let's just say that it's stuck between two other servers without any breathing room ​:this_is_fine:​

A while ago, I've set-up the ai.robots.txt nginx config to return a 403 to all AI crawlers based on their user-agent.
I thought it'd be ineffective as these crawlers are known to switch user-agents when being blocked, but it worked fine and significantly dropped the amount of requests I was getting (once the crawlers had emptied their queues).
Now that I'm switching to something else than nginx, I haven't yet taken the time to look at solutions similar to the one I just described, but as I migrated my git server from my old infra to my new one, I quickly checked the logs out of curiosity...
And crawlers were already hammering my server, opening blames and downloading my code left and right...
It was only back open for less than a minute! ​:dumpster_fire:​

For more context, I'm talking about the PTU sound, as heard (and explained) in this YouTube short ​:kirby_headphones:​

Weeeeeeeeeeee oooooo woof woof woof woof woof

"LinkedIn is replacing Kafka"

Shout-out to @mikrotik@mikrotik.social RouterOS built-in Packet Sniffer ​:clap_clap:​ incredible built-in tool for debugging networking and firewall issues ​:epic_handshake:​

"As your AI-enhanced washing machine isn't supported anymore, we remotely applied a firmware update to turn it into a crypto miner."
#microfiction

There are a lot of cursed computers things but this is the most cursed computers thing

https://benjojo.co.uk/u/benjojo/h/h4N78m1PjXYsYfzkGV