Major PG slow down after forced service stop by Railway.
iiskakov
PROOP

a year ago

Hi there! I haven't timely paid you, and my Railway account activity was stopped.

I then paid my subscription, and had to redeploy the services (after they were stopped). However they became VERY slow for some reason (the code has not changed, literally same commit). I'm not sure what's happening.

Postgres log:

2025-03-09 17:47:31.783 UTC [36] LOG: checkpoint complete: wrote 462 buffers (2.8%); 0 WAL file(s) added, 0 removed, 1 recycled; write=46.227 s, sync=0.006 s, total=46.242 s; sync files=216, longest=0.002 s, average=0.001 s; distance=3661 kB, estimate=3661 kB

Note the 46s checkpoint time.

P.s. My database is rather small

Solved

10 Replies

a year ago

A project id is relevant here. please provide one


iiskakov
PROOP

a year ago

7e603fcd-3ad0-4cf7-8c8c-2492422747af


iiskakov
PROOP

a year ago

🥺


iiskakov
PROOP

a year ago

I suspect this might be something with railway infra


iiskakov
PROOP

a year ago

Folks? 😦


iiskakov
PROOP

a year ago

I've created a Duplicate of postgres, and it has the same issue:

2025-03-10 06:52:56.172 UTC [104] LOG: checkpoint complete: wrote 222 buffers (1.4%); 0 WAL file(s) added, 0 removed, 0 recycled; write=21.975 s, sync=0.004 s, total=21.983 s; sync files=81, longest=0.002 s, average=0.001 s; distance=2452 kB, estimate=2452 kB

write takes 21.975s for a small checkpoint


a year ago

looking at the wal time is not an indication of performance in the slightest.

please make sure you have all your services within a project in the same region, zone, and datacenter.


iiskakov
PROOP

a year ago

Thank you, Brody!

Changing from Amsterdam (Metal) to Amsterdam helped! I actually don't ever remember switching to metal,and frankly don't really know what it's for. I'll look it up!

Thanks again!


a year ago

no problem!


a year ago

!s


Status changed to Solved brody about 1 year ago


Loading...