connection failed with schedule_in_database

I've used schedule_in_database() to schedule a job. My job looks like this:

```
jobid | schedule | command | nodename | nodeport | database | username | active | jobname
-------+-----------+--------------------------------+-----------+----------+----------+----------+--------+-----------------------
1 | * * * * * | select feedlot.update_caches() | localhost | 5432 | feedlot | postgres | t | feedlot_update_caches
(1 row)
```
However, the jobs always fail with a `return_message` of "connection failed":

```
postgres=# table job_run_details;
jobid | runid | job_pid | database | username | command | status | return_message | start_time | end_time
-------+-------+---------+----------+----------+--------------------------------+--------+-------------------+------------+----------
1 | 2844 | | feedlot | postgres | select feedlot.update_caches() | failed | connection failed | |
1 | 2845 | | feedlot | postgres | select feedlot.update_caches() | failed | connection failed | |
```

I don't know what I am doing wrong or how to debug it. Logs just say the same thing:

```
2024-01-08 17:06:00.002 UTC [11328] LOG: cron job 1 starting: select feedlot.update_caches()
2024-01-08 17:06:00.009 UTC [11328] LOG: cron job 1 connection failed
```

The `postgres` user is of course a superuser and has access to everything. Any ideas?