2 years ago
I'm running about 1000-1200 raw queries (logged through Prisma) in a few seconds and am getting wait times of 20-30 seconds for something that takes 1ms with pg's EXPLAIN ANALYZE.
I don't see anything spike in the Railway panels for our Postgres instance either. The only one that spikes is outbound, which hit about 120mb. I don't think that's unreasonable though and shouldn't be causing these long load times and backup.
It could be:
network latency (maybe we are querying too much info?)
a Prisma issue
a Postgres issue
a Railway config issue with Prisma and/or Postgres
Has anyone run into anything similar?
Am I using Prisma incorrectly? I've barely changed any of the default settings.
4 Replies
2 years ago
I don't use Prisma, just Postgres + Python, and I'm also experiencing a lot of unexplained latency at non-uniform intervals. I'm not sure how to debug this either. I set up a staging server on another provider and I'm not seeing it there. Still trying to reproduce this issue.
2 years ago
Yes! I just tried using the private database URL on the deployed instance.
I don't think 1000 queries should warrant an issue like this that requires network optimization, but please let me know if I am wrong!
@rc
2 years ago
I don't use Prisma, just Postgres + Python, and I'm also experiencing a lot of unexplained latency at non-uniform intervals. I'm not sure how to debug this either. I set up a staging server on another provider and I'm not seeing it there. Still trying to reproduce this issue.
Let me know if you figure out what it may be! what other provider did you use? Would like to test