tobi
tobi
@NyaaaWhatsUpDoc should we still consider this issue open? Or should we close it? I don't think we're having too much trouble with our config stuff these days.
> I'm not exactly sure what m2m means maybe @tsmethurst can fill in on that many-to-many :) It's a type of relation, like one to many, many to one
No rush, hope you feel better soon!
Heya, I'm gonna close this for now, but let us know if you want to reopen it again, it's no problem :)
Thank you for the detailed report! Much appreciated! It's sort-of-known that ram use spikes when a post gets boosted or replied to by an account with 1k or more followers,...
> SwiftOnSecurity with 28,000 followers just boosted one of my posts and GtS has been bouncing up and down like a trampoline getting repeatedly OOMkilled Oh shit :D > I'm...
Thanks for gathering this data :) > A simpler approach might be adding HTTP middleware that queues, or simply rejects, (expensive?) requests over a (configurable) parallelisation limit So throttling basically?...
> Do we have traffic generation scripts for this sort of testing already? No not yet; we haven't done performance or load testing beyond informal stuff. > the second is...
Mm might be worth a try. We briefly tried caching some json responses in memory in GtS last year or so, but back then our cache implementation (read: something I...
mm, it shouldn't be the logging, likely it's the amounts of accounts your instance has to go and dereference when it suddenly gets all those requests to view the status...