Skip to content
PostgreSQL checkpoint write latency in europe-west4 (HOBBY)
oguzcanuludag
OP

a month ago

Hi, our Postgres service is showing slow checkpoint writes, and the backend sees intermittent read timeouts. The database itself is idle.

Setup

Service: Postgres, environment production

Plan: HOBBY, 1 replica, region europe-west4 (the backend is in the same region)

PostgreSQL version: PostgreSQL version: 17.11

Checkpoint logs (UTC, 2026-10-16)

10:59:17 checkpoint complete: wrote 23 buffers; write=2.213 s; sync=0.098 s; total=2.377 s

11:04:21 checkpoint complete: wrote 53 buffers; write=5.241 s; sync=0.041 s; total=5.487 s

11:09:16 checkpoint complete: wrote 10 buffers; write=0.921 s; sync=0.027 s; total=1.029 s

11:14:19 checkpoint complete: wrote 5 buffers; write=0.419 s; sync=0.057 s; total=0.617 s

11:19:21 checkpoint complete: wrote 57 buffers; write=5.622 s; sync=0.033 s; total=5.720 s

11:29:18 checkpoint complete: wrote 8 buffers; write=0.706 s; sync=0.011 s; total=0.774 s

11:34:23 checkpoint complete: wrote 76 buffers; write=7.540 s; sync=0.047 s; total=7.626 s

11:39:19 checkpoint complete: wrote 8 buffers; write=0.929 s; sync=0.362 s; total=3.674 s

Write time is about 0.1 s per buffer, which points to per-I/O latency rather than throughput.

Resource usage in the same window: Postgres CPU around 0.05 vCPU, memory around 120 MB, disk usage about 130 MB and flat, network negligible. The workload is light.

Questions

  1. Is the host for this service currently experiencing elevated disk write latency?
  2. If so, is there an expected time to recovery?
  3. Is there anything we can change on our side without moving the database, or do you recommend a migration?

We have not restarted or redeployed the service. Thank you.

Solved

1 Reply

Railway
BOT

a month ago

Your Postgres volume runs with the standard limits of 3,000 IOPS and 70 MB/s, and the write volume in your checkpoint logs is far below both, so these slow writes are not from reaching those limits.

We do not have a setting change on your side that we can point to as a fix, and we are not recommending moving the database based on what we can see right now.


Status changed to Awaiting User Response Railway • about 1 month ago


Railway
BOT

23 days ago

This thread has been marked as solved automatically due to a lack of recent activity. Please re-open this thread or create a new one if you require further assistance. Thank you!

Status changed to Solved Railway • 23 days ago


Welcome!

Sign in to your Railway account to join the conversation.

Loading...