All I am doing is irritating people.
In the Lemmyverse: People seem to enjoy upgrading hardware to huge numbers of cores, adding big-brand front-end commercial DDOS protection, but do not like !lemmyperformance@lemmy.ml discussion on Lemmy itself (eat your own dog-food, developers is not a mantra, as I’ve said here before).
Each new local comment and post on a local server updates 1800 rows in site_aggregates, but there seems no urgency to get the fix out first thing Monday- and stop every large Lemmy server from crashing so frequently. I put “emergency” in the pull request title, leadership edited it out. That says it all. EVERY ONE of the major servers is crashing on my personal visits all weekend, Monday, now Tuesday.
I bring up a direct topic of engineering deletes for scale and concurrency and get push-back from the players on GitHub that that’s off-topic to the delete crashing problem, and I have to lecture them on autism mental differences, which they can’t wait to be irritated by. What a shock, people who find autistic thinking irritating and different from what they think.
It’s the lack of testing and sharing the backend data directly that nobody even noticed these things (site_aggregates has 1700 rows, all getting increments for every new post and comment). And push back on every code that doesn’t pass beautification tests or has an unused testing function in it - it’s like a massive corporation where people only work 9 to 5 and find every excuse to say “not my problem” when it comes to servers crashing every hour.
So the first hour of work on GitHub, sanitizing emails and enabling gzip compression seems to be a priority. Crashing servers still not a project priority. but the tiny number of people who use an e-mail feature that could be disabled during the server-crashing crisis… is getting the attention. Back on June 1 when the servers were crashing I was confused, but that’s kind of the point, is’n’t it?
Social hazing still the best answer I can come up with, not just developers, but every newcomer to the project… the front page of the GitHub bragging “full delete” and “high performance” - lays the foundation for the entire project, even typical end-users, to go down that path. People rarely discuss why hazing evolves in some business and especially military and school structures, and it shows up across a variety of demographics. It isn’t a money motivation, it’s a social focus in most cases. For me, it’s a new experience to witness it see it utilized on the scale of an entire multi-operator organization - all the newcomers and established server operators. lemmy.ca took it in their own hands to track down the performance bug that had sat on GitHub for over 30 days… that server has been around since December 28, 2020 at least - so they have a lot more experience with the priorities of developers on performance - and got into detailing the SQL crashes directly. Hazing is really something to witness.