PushProx is a client and proxy that allows transversing of NAT and other similar network topologies by Prometheus, while still following the pull model.
While this is reasonably robust in practice, this is a work in progress.
First build the proxy and client:
git clone https://github.com/prometheus-community/pushprox.git
cd pushprox
make build
Run the proxy somewhere both Prometheus and the clients can get to:
./pushprox-proxy
On every target machine run the client, pointing it at the proxy:
./pushprox-client --proxy-url=http://proxy:8080/
In Prometheus, use the proxy as a proxy_url
:
scrape_configs:
- job_name: node
proxy_url: http://proxy:8080/
static_configs:
- targets: ['client:9100'] # Presuming the FQDN of the client is "client".
If the target must be scraped over SSL/TLS, add:
params:
_scheme: [https]
rather than the usual scheme: https
. Only the default scheme: http
works with the proxy,
so this workaround is required.
The /clients
endpoint will return a list of all registered clients in the format
used by file_sd_configs
. You could use wget in a cronjob to put it somewhere
file_sd_configs can read and then then relabel as needed.
Clients perform scrapes in a network environment that's not directly accessible by Prometheus. The Proxy is accessible by both the Clients and Prometheus. Each client is identified by its fqdn.
For example, the following sequence is performed when Prometheus scrapes target fqdn-x
via PushProx.
First, a Client polls the Proxy for scrape requests, and includes its fqdn in the poll (1).
The Proxy does not respond yet.
Next, Prometheus tries to scrape the target with hostname fqdn-x
via the Proxy (2).
Using the fqdn received in (1), the Proxy now routes the scrape to the correct Client: the scrape request is in the response body of the poll (3).
This scrape request is executed by the client (4), the response containing metrics (5) is posted to the Proxy (6).
On its turn, the Proxy returns this to Prometheus (7) as a reponse to the initial scrape of (2).
PushProx passes all HTTP headers transparently, features like compression and accept encoding are up to the scraping Prometheus server.
There is no authentication or authorisation included, a reverse proxy can be put in front though to add these.
Running the client allows those with access to the proxy or the client to access all network services on the machine hosting the client.