Basic debugging:
Also make sure you are always using health_check
Specific debugging:
2025-06-19 16:44:31.934: main pid 6248: LOG: reading status file: 1 th backend is set to up status
This would seem to indicate it's reading cached info about the state of the node, so you may need to remove the pgpool_status file. You can also tell pgpool to ignore it on startup via the -D or --discard-status flag.
Hope that helps,
Greg