docker login fails
docker login fails
Your Docker Hub token has likely expired or was mistyped. Confirm it
with RELAI and re-run:
app-migrations exits with an error
app-migrations exits with an error
Check its logs:This is usually a database credential mismatch — confirm
DB_NAME,
DB_USER, and DB_PASSWORD in .env match what the db container was
created with. If you changed DB_PASSWORD after the first run, the
MySQL data volume still has the old password; either reset the
mysql_data volume (destroys existing data) or update the password
inside MySQL to match.A port is already in use on the host
A port is already in use on the host
Something else on the host is bound to
8889, 5173, DB_PORT, or
REDIS_PORT. Either stop the conflicting process, or change the
relevant port variable in .env and re-run
docker compose up -d --remove-orphans.The dashboard can't reach the backend
The dashboard can't reach the backend
Check that
SERVER_BASE_URL in .env matches how the dashboard should
reach the backend (e.g. if you’ve put a reverse proxy or a different
hostname in front of it), then restart the dashboard service:relai init fails with "The model proxy configuration is invalid"
relai init fails with "The model proxy configuration is invalid"
The backend is missing a required model proxy setting, usually
RELAI_MODEL_PROXY_MODEL or RELAI_MODEL_PROXY_CONTEXT_WINDOW_TOKENS.
Backend 1.72.0 and later require both. Add them from your provider’s block
in Configuring model provider,
then recreate the backend services:relai init fails with "invalid_tool_schema"
relai init fails with "invalid_tool_schema"
The proxy refused a tool definition that the provider would reject. This
happens only when
RELAI_MODEL_PROXY_WIRE_API="chat_completions" is set.
Update the relai CLI (relai update), or use responses or auto.relai init reports that the model's response was cut off
relai init reports that the model's response was cut off
The model reached its output budget before finishing. With
RELAI_MODEL_PROXY_WIRE_API="responses", raise
RELAI_MODEL_PROXY_MAX_OUTPUT_TOKENS (the default is 32000), then recreate
the backend services.Something else looks wrong
Something else looks wrong
Check the logs for the specific service:If you’re still stuck, reach out to RELAI with the output.