Querying backend/upstream server status

[Hungpo DU]

  Best regards.


On Wed, Jul 14, 2010 at 10:38 PM, Hungpo DU <alecdu at gmail.com> wrote:

>
>
>
> On Wed, Jul 14, 2010 at 7:01 PM, Srebrenko Šehić <ssehic at gmail.com> wrote:
>
>> On Wed, Jul 14, 2010 at 4:15 AM, Weibin Yao <nbubingo at gmail.com> wrote:
>>
>> Hi,
>>
>> > See this module as an example:
>> > http://github.com/cep21/healthcheck_nginx_upstreams
>>
>> Yeah. I had a look at that, but that module does an active health
>> check. All I want is to query the current status as seen by nginx'
>> workers themselves. I do NOT want to send any additional requests to
>> the upstream.
>> Any pointers?
>>
>> I don't think it's possible under current nginx architecture.  Worker
> processes are quite independent, and nginx don't have mechanism like
> Apache's scoreborad, where every worker writes its status on. At least, you
> can't get the global information about what every process are doing.
> But, you may get upstream status about the worker_process who is handling
> your "/monitor" request by checking every requests the worker_process is
> handling? Just a thought.
>
> Yes, some idea hits me. Maybe, what ngx_http_stub_status_module does is
what you want. It does have gobal information about all the workers. Every
worker writes it's connections and requests information into a piece of
shared memory. But, it may requires a lot of work to show you what you want.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://nginx.org/pipermail/nginx-devel/attachments/20100714/49c51d52/attachment.html>