x86pup

Maybe it'll help if I explain my usecase here too. I currently maintain [hardened_malloc-rs](https://github.com/girlbossceo/hardened_malloc-rs). I made this because I would like to integrate hardened_malloc into Rust crates without relying on...

Ran into this issue too with postgresql via UNIX socket + S3, unfortunately destroyed my caches before stumbling upon this issue and the PR fix. I'm going to safely assume...

Running into this issue too with my CI. Reverse proxy is Caddy, attic is using postgresql via UNIX sockets + S3 via garage. Occasionally it'll cause it to just stop...

I think your Caddyfile seems to be giving me better results. I re-ran a CI job that pulls from attic and I don't see any failures now.

Unfortunately still getting HTTP/2 errors: https://gitlab.com/conduwuit/conduwuit/-/jobs/6952157628#L440 I'll try to re-read your comment again and see if I missed something.

Had the GitLab mirror private for some reason (how were other people able to star it in the past?) I'll re-run CI on the GItLab with trace logs again

https://girlboss.ceo/~strawberry/pb/qa2A Nothing really sticks out when the HTTP/2 errors happen except for one of these: ``` May 27 20:03:15 TPYEXECL.local atticd[2867619]: 2024-05-28T00:03:15.893193Z DEBUG hyper::proto::h2: send body user stream error: error...

Oh, yes I was. I knew this would be a footgun at some point :< really wish there was a proper way to migrate over from local to S3, I...

Perhaps these HTTP/2 errors are actually just database errors not being shown to the user at the end of the day considering [oluceps said the HTTP/2 errors disappeared after deleting...

> for now you could make an one-off script that changes all remote_file and remote_file_id in a loop and it should work. This has worked! (and I am very impressed...