Feature requestED: monitoring Nginx from the outside

[François Battail]

Rt Ibmer <rtibmx at ...> writes:

> So let's say that I want to write a shell script that would continuously
monitor nginx to know  when an upstream server went down, and if so, which
upstream server it is (IP or name).
With the approach you are proposing, can you explain how I would go about this?
What would my shell script look like? It sounds like I would still need to
modify nginx or write a module to monitor for this and tell nginx to report such
a condition?  What would be involved for that?

Well, *if* module "ngx_http_upstream_round_robin" reports up and down events it
would be something like this:

upstream-192.168.0.1:0
upstream-192.168.0.2:1
...

The shared memory is just filled with text, so it is up to the module to name
the variable, in that case using the ip or dns name of servers in the name seems
obvious.

At this time nothing is really done, it's just a suggestion, I believe it could
be interesting for Nginx to have such feature but without any compromise. It
means code should be portable, not bloated *and* accepted by Nginx community
*and* accepted by Igor.

Why? Because it's not a standalone module but rather an API, so other modules
with valuable data for monitoring need to be modified. But it's not so easy.
It's not a matter of simply writing code but to evaluate all implications. Just
for example: security, portability, support and side effects.

As there's an ecosystem based on Nginx a modification of this kind may affect
other softwares (it should not be the case).

So, it takes time to discuss about some things:

- is it really needed?
- what is the impact?
- does it broke something which was working?
- is it blocking the evolution of the software?
- is it clean and maintainable?
...

It is up to this list to answer these questions or to add another questions (or
suggestions).

Best regards.