#computersarehard — Public Fediverse posts

There are some days when you feel all powerful. Building entire services. Squashing production bugs. Solving riddles.

Other days you spend the entire day trying to pair a lightbulb.

#iot #coderlife #computersarehard

I experimented with adding some #MRF policies in #akkoma cause I just wanted to play around with them. I wasn’t sure what some of them do and they’re not all very well described in the docs or in admin-fe so I decided to look at the #container #logs. Being smart I just ran docker logs -f akkoma_akkoma_1 which logically blasted me with a wall of text.

I forgot I left that running and then a a bit later I saw that the #VPS, normally doing on average no more than 10-20% CPU utilization, sub 10MBps disk throughput, sub 1MBps network traffic, was suddenly frying itself at ~250% CPU, ~30MBps disk ~15MBps network for the past 10 minutes or so (roughly when I enabled those MRF policies).

Being “super smart™“ I immediately decided I’ve enabled some policy that’s doing something super complicated and massively intensive, mildly panicked that I’m going to kill my instance and went on to disable most of them. By this point you will be very surprised to know it did not help.

Continuing my streak of genius I then proceeded to stop the (still running) docker logs -f akkoma_akkoma_1 command and start another one, but adding --since 10m to try and see which of the policies was the culprit. Few seconds later the graphs now showed CPU flatlining at 400%.

narrator voice: And that’s when he knew he was dumb AF.

Hitting ctrl-c on that docker logs wasn’t enough as apparently containerd keeps going crazy under the hood even if no one’s actually consuming those logs. 5 minutes later I just rebooted the VPS and everything was back to normal after that.

Sp yeah… #computersAreHard, but also I’m kind of a simpleton myself really…