Hi there,

Sometimes an external data source may become unavailable and we need to set
a far more aggressive timeout on our outgoing requests, and I'd like to
base that decision on the uwsgi.listen_queue() value.

But from a production environment running uwsgi 2.0.12, it seems that the
numbers returned from listen_queue() doesn't match what uwsgitop says once
in a while - sometimes it gets stuck on a random number like 62, other
times, it sticks to zero while queue is overloaded at 101.

The stuck value varies per worker, it seems like when the number is moving
around like up or down, it broadcasts to most workers correctly, but when
it reaches 101 or 0, some clients get stuck at the last known (wrong) value.


As I'm not too familiar with C, I don't think I can track the issue down
myself... Any help with this would be appreciated. For now, I'm thinking
about modding uwsgitop to publish the correct value once per second to the
cache.


Kind regards,

-- 
Chi Ho Kwok
Digibites Technology
[email protected]
_______________________________________________
uWSGI mailing list
[email protected]
http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi

Reply via email to