notify: empty or broken HTTP response

[antonioparisi]

Hi guys,

every ~2 days the on_publish notification stops working and the logs is this:

2017/02/21 14:27:15 [info] 264#0: netcall: client send timed out (110: Connection timed out)
2017/02/21 14:27:15 [info] 264#0: *699071 notify: empty or broken HTTP response, client: x.x.x.x, server: 0.0.0.0:1935
2017/02/21 14:27:15 [info] 264#0: *699071 disconnect, client: x.x.x.x, server: 0.0.0.0:1935
2017/02/21 14:27:15 [info] 264#0: *699071 deleteStream, client: x.x.x.x, server: 0.0.0.0:1935
2017/02/21 14:27:15 [info] 264#0: netcall: client send timed out (110: Connection timed out)
2017/02/21 14:27:15 [info] 264#0: *699072 notify: empty or broken HTTP response, client: x.x.x.x, server: 0.0.0.0:1935
2017/02/21 14:27:15 [info] 264#0: *699072 disconnect, client: x.x.x.x, server: 0.0.0.0:1935
2017/02/21 14:27:15 [info] 264#0: *699072 deleteStream, client: x.x.x.x, server: 0.0.0.0:1935

the notify in question is a on_publish that calls a URL (and the URL is always the same). The URL is kind of similar:

on_publish http://URL/auth/token;

and then it appends parameters that comes from the request like i.e. ?token=foobar&....

So as I said it works normally for the first 2 days, then it start bombing out with the above logs. Any idea?

Thanks

[antonioparisi]

@sergey-dryabzhinsky any input on this?

[michaelhelmick]

Hi, I've noticed this behavior too and would like to avoid doing a cron to nginx -s reload

@sergey-dryabzhinsky I'm not sure how often you contribute to this repository anymore, but this is a thorn :( any input is appreciated <3