infra_portal endpoint settings and consequences

[VadimJCA]

Pigsty v3.7.0 on Rocky Linux 9.7

What would be the proper way to fix homepage and get prometheus to talk to alertmanager if changing endpoint ports in infra_portal and reloading nginx is not enough?

---

When configuring endpoints for prometheus and alertmanager in pigsty.yml: all: children: vars: infra_portal I initially used ports 9090 for prometheus and 9093 for alertmanager (from the earlier config). After successfull installation all INFRA and PGSQL dashboards were working fine. Later I noticed that URLs at my pigsty's homepage lead nowhere. Except for the homepage and grafana - those were fine. But prometheus and alertmanager pages didn't open.
I did some digging:
sudo systemctl status prometheus
and found out that prometheus Web UI listens on the other port - 9058
/usr/bin/prometheus --config.file=/etc/prometheus/prometheus.yml --web.listen-address=:9058 "--web.page-title=Pigsty Metrics" --storage.tsdb.path=/data/prometheus/data --storage.tsdb.retention.time=15d
And it also has some troubles with alertmanager

time=2025-12-29T12:18:04.703+05:00 level=ERROR source=manager.go:495 msg="Error sending alerts" component=notifier alertmanager=http://10.0.0.111:9093/api/v2/alerts count=1 err="Pos>

So I went and changed endpoint ports to those 9058 ad 9093 respectively in pigsty.yml and ran ./infra.yml -t nginx to reload nginx. URLs on homepage came alive, but all grafana dashboards went blank.
I've struggled with rolling back, until I reran:

./infra.yml
./etcd.yml
./node.yml -t node_exporter
./pgsql.yml -t monitor,register

And I've got prometheus and grafana back online! Except for the prometheus error when sending alerts to alertmanager and half-dead homepage.

Please advice how to get everything in order.