As mentioned (very) briefly in other posts, I run Home Assistant as the ‘control hub’ for all my “smart” devices in my home. I originally ran it via their ‘HASSOS’ Docker image but was never really happy with it. So when I most recently rebuilt my NUC, I decided to give this Podman thing a look. And so far, I seem to like it better. I’m still not entirely sold on this whole container bullshit, but whatever ;)
About a year ago (Sep 2019 to be precise) I decided to end my Raspberry Pi experiment and begin a new experiment with an Intel NUC. It’s not that the Pi is incapable or anything. I really like the platform and will probably find something else to use it for. But my computing needs/desires had changed and I was looking at having a small fleet of them (I already had 2 and was contemplating more) and I really didn’t want to go that route. So I looked around, did some research, saved up some money and ended up with:
We’ve had a small flurry of customers asking about tuning their OS for the best PostgreSQL performance. While the answer to this question is always ‘that depends on your hardware and workload’ and involves a lot of iteration between changing a setting and benchmarking, I thought I’d take a moment to point out that once you do manage to dial-in the settings, you should be writing a profile and deploying to your systems for tuned to make use of. Please, for the love of $diety, stop editing sysctl.conf and friends!
UPDATE: My reasoning was incorrect below. It wasn’t the moving of some of the lock code to C that caused the issue. It was moving -D_POSIX_C_SOURCE up to the Makefile that caused the problem. The solution below is still the same though.