I am seeing normal response times as of today. Yesterday was hit or miss but the latter half of the day seemed substantially better.
Fingers crossed!
I am seeing normal response times as of today. Yesterday was hit or miss but the latter half of the day seemed substantially better.
Fingers crossed!
We have switched to type = "requests"
and we are still seeing very inconsistent request times.
We have also tried different VM sizes (shared vs dedicated) and the results seem to be random between deploys, leading us to think there might be a noisy neighbor issue?
The range on a very static endpoint can range from 100ms to 1200ms from the same machine/location.
My instincts lead me to think this is a PG connection issue? We are only running app servers and pg servers in a single region (ORD).
We have also noticed that when running fly in only a single region to prevent DB connection issues we really see slow connections from anywhere else in the world. Compared to running only on heroku iad, this never seemed to be the case, but on fly we are seeing noticeably slower requests when far from the hosting region, anything we can do to improve this?
(I should add this is all node/prisma)
Thanks in advance!