>postgres instability
>Misskey
I'm a tad surprised. I'm not experienced with it but you never hear of Misskey servers keeling over from what I can only describe as database mutational load (see: Pleromer). Did you do anything particular with the postgres confs?
