10 days ago
I just deployed the template without any issues, and all services are green and I don't see any errors in any of the logstreams.
I'm now trying to connect to the primary entrypoint via the TCP proxy that was pre-configured on haproxy, using pgadmin, but I only get:
connection failed: connection to server at "xx.xx.xx.xx", port xx failed: server closed the connection unexpectedly
This probably means the server terminated abnormally
before or while processing the request.Surely missing something obvious?
4 Replies
10 days ago
When I attempt connecting, the expected log line does appear in haproxy logs:
Connect from ::ffff:100.64.0.4:42478 to ::ffff:10.250.19.72:5432 (postgresql_primary/TCP)
Also, accessing the DB via the haproxy "Database" tab from Railway UI fails to connect as well. The three DB nodes can all be accessed nominally
10 days ago
Alright things are getting weirder: connecting to the replica via port 5433 (changed the TCP proxy on haproxy to point to 5433) immediately allows connections. Only when the TCP proxy is pointed at 5432 it fails
10 days ago
The haproxy dashboard shows a number of connection problems (see attachment).
I have a feeling this may be related to our environment being IPV6-only (created before IPV4 support was released)...?
Attachments
10 days ago
I set PATRONI_RESTAPI_LISTEN=[::]:8008 on all pg nodes - and now everything appears to be working properly. Weird!