Monitoring the health of NPM

[Australian]

Hi all,

Looking for some help with monitoring Nginx Proxy Manager after some odd behaviour.

I am fine with troubleshooting the Server & Platform that NPM sits on but i am not as experienced with NPM itself and was hoping someone could give me a steer.

There have been a few instances where the server becomes unreachable but yet still kind of works. I cannot browse to the NPM Dashboard, cant RDP or Teamviewer onto the server and cant even open the Console on VMware but CAN SSH onto it and the proxied hosts are... proxied fine. A magic reboot fixes it.

So the Network Proxy Manager is running on Ubuntu Server 24.04 which is a VM running on VMware ESXi 8
Connectivity via OpenSSH, RDP and also Teamviewer as a backup way of getting onto my network when i am outside and if for whatever reason OpenVPN is not available.

I am using PRTG for Monitoring.

Any ideas on what i should be monitoring within NPM ??

[tigran-sargsyan-w]

I would monitor this in layers rather than using only an "NPM is up/down" check.

One important clue in your description is:

NPM dashboard unreachable
RDP/TeamViewer unreachable
VMware console unavailable
SSH still works
proxied hosts still work

If the proxied hosts are still responding, the Nginx data plane is probably still alive.

And if even the VMware console has problems, I would not assume NPM itself is the root cause.

For PRTG I would monitor at least these independently.

VM / host

CPU, memory/swap, disk space, disk latency and load.

On Ubuntu:

free -h
df -h
df -i
uptime

Docker

Container state and restart count:

docker compose ps
docker inspect \
  -f '{{.State.Status}} restarts={{.RestartCount}}' \
  <npm-container>

Nginx configuration/process

docker exec <npm-container> nginx -t

NPM management plane

Probe the management UI/API separately from the actual proxy traffic.

For example:

curl -fsS http://127.0.0.1:81/ >/dev/null

(adjust depending on how port 81 is exposed).

Real proxy traffic

Have PRTG request one or more actual Proxy Hosts:

curl -fsS https://service.example.com/ >/dev/null

This distinction is useful.

If:

port 81 DOWN
proxy hosts UP

then Nginx is serving traffic but the management/backend side has a problem.

If:

VMware console + several guest services DOWN

at the same time, I would investigate ESXi/VM resources before blaming NPM.

I would also monitor certificate expiry separately, because the containers can be perfectly healthy while an HTTPS service is about to fail due to an expired certificate.