<@U04S8PLJWMT> - i am getting a gateway 504 error ...
# ask-ai
a
@kapa.ai - i am getting a gateway 504 error for my self hosted that began last week between airbyte and snowflake. i jsut focred the update to 4.0.30 and no idea why this issue is occuring. An unknown error occurred. (HTTP 504)
k
A 504 in a self‑hosted Airbyte → Snowflake setup generally means the UI/API timed out waiting for the backend to respond, not necessarily that Snowflake itself is down. From the cases in the knowledge you shared, there are three main patterns and workarounds: 1. Reverse proxy / basic‑auth proxy timeout When Airbyte is fronted by nginx or the built‑in basic‑auth proxy, long‑running operations (schema discovery, syncs, etc.) can hit the default timeout and return 504. Increasing the proxy timeout is the suggested fix. For Docker deployments this is controlled via
BASIC_AUTH_PROXY_TIMEOUT
in the
.env
file; support explicitly recommended raising it, for example to 600 seconds (10 min). [[Airbyte proxy 504](https://discuss.airbyte.io/t/3305)] > Action: > - If you’re using the default Docker setup, edit your
.env
and set: > -
BASIC_AUTH_PROXY_TIMEOUT=600
(or higher if needed) > - Restart Airbyte (
docker-compose down && docker-compose up -d
or equivalent). 2. Insufficient resources on the Airbyte host Several users hit 504 and “UI unreachable” when the instance ran out of CPU/RAM or disk; upgrading the instance size resolved it. [[Instance unreachable](https://discuss.airbyte.io/t/3037); [Zuora schema 504](https://discuss.airbyte.io/t/2976)] > Action: > - Check CPU/memory usage on the Airbyte host during the failing sync. > - If it’s high or the worker pods are stuck/pending, allocate more RAM/CPU or a larger instance and retry. 3. Version‑specific / deployment‑specific issues - Some older versions (around
0.40.x
) had upgrade / bootstrap problems that were often addressed by upgrading again to a newer patch (e.g.
0.41.0+
). [[Upgrade 0.40.x→0.41](https://discuss.airbyte.io/t/3916); [Won’t start 0.40.30](https://discuss.airbyte.io/t/3843)] - There are also reports in more recent Kubernetes/abctl installs where
504
appears when the server loses access to underlying infrastructure (permissions, networking), sometimes resolved by downgrading or changing chart versions. [[abctl 504, downgrade](https://github.com/airbytehq/airbyte/issues/62463)] > Action: > - If you’re on
0.40.30
and not pinned for a reason, consider upgrading to the latest stable version supported by your deployment method and retry. > - If you deployed via
abctl
or Helm, ensure all Airbyte pods are healthy and can reach Snowflake (no pods stuck in
Pending
or
CrashLoopBackOff
). [[Syncs stuck & 504](https://discuss.airbyte.io/t/2394)] Because the knowledge you provided doesn’t include a Snowflake‑specific 504 root cause beyond timeouts and general resource / deployment health, I can’t point to a confirmed bug specific to Snowflake itself. To move forward, based on the docs above I’d suggest, in order: 1. Increase
BASIC_AUTH_PROXY_TIMEOUT
(or your nginx
proxy_read_timeout
if you run your own nginx) and restart. 2. Verify the host has sufficient CPU/RAM and that syncs aren’t stalling due to resource exhaustion. 3. If possible, upgrade from
0.40.30
to a newer stable version and retry the same sync. 4. If the 504 persists and logs don’t show Snowflake SQL errors (permissions / database, etc.), share the server and job logs for the failing connection in a new forum thread so the team can inspect further; that’s how similar cases have been handled. [[Snowflake connection stopped](https://discuss.airbyte.io/t/1178)] If you can share: Airbyte deployment type (Docker vs K8s/abctl), exact version, and whether you have any proxy in front (nginx, load balancer), I can help narrow which of these is most likely.