Unmanaged Postgres, single node, repmgr fails on boot

[ronaldorp]

Out of the blue Im getting this error and the machine won’t start:

• repmgr tries to connect to 48ed1eeb303078.vm.my-app-db.internal from inside the same VM and that hostname doesn’t resolve.

• Log line: could not translate host name “48ed1eeb303078.vm.my-app-db.internal” to address: No address associated with hostname.

Does anyone have an idea on what’s happening?

Any help would be much appreciated

[mayailurus]

Hm… Is 48ed1eeb303078 definitely the Machine’s own ID? Possibly you had multiple nodes in the past, and this is a remnant of that older configuration, .

If this persists, then I would try the volume-forking technique (with explicit volume ID) and see whether that new Machine boots successfully…

---

Aside: This might be a good time to reassess whether single-Machine Legacy Postgres is really the right fit for the data you have in there. The next glitch like this could very well wipe out the data itself, …

@john-fly

But many users use single-machine Fly PG for production! Please don’t do this – it’s not a supported configuration! I’ve had to break the news to several users recently that their data was lost/may be lost when a host goes offline for a (suspected) hardware failure, taking the data with it. Always run a Fly PG cluster with 3 or more nodes in production.

[ronaldorp]

Thanks folks. It is a stage environment, so no issue with the single node unmanaged PG, but I was (am) concerned that could happen in prod. Thanks for pointing it out though.

48ed1eeb303078 is a machine I own.

After wasting a afternoon playing with Postgres user authentication for `repmgr`, the solution was to add it on my db.toml:

[env]
#Force self-connection via localhost (postgres-flex expects this for single-node)
FLY_PRIVATE_IP = ‘127.0.0.1’

Thanks for your help