I'm now fairly convinced that it was a
I'm now fairly convinced that it was a Hyperdrive issue. As a hot-fix, I just deployed with a raw postgresjs client (no hyperdrive) and the errors stopped
5 Replies
creating a thread with screenshots
can see that the Errors in the monitoring tab spikes. i couldn't see any logs in the worker observabiltiy dashboard, but i was seeing tons of client disconnected errors

started seeing impact 3/15/2025 4:25PM PST

I deployed a change where I just instantiated a regular postgresjs client at 6:22PM PST and the errors mostly went away

this is what i see in my hyperdrive dashboard. note that i had caching disabled.
there's a big spike around 4:32PM PST to 1.2K QPM.
i tried enabling caching around 5:38PM PST to see if it would help, but since it didn't i disabled it.

