[uWSGI] possible bug? if the first and only request is a websocket request, harakiri is triggered

Roberto De Ioris roberto at unbit.it
Mon Apr 7 06:38:57 CEST 2014


> Sorry, I forgot to post the answer the day I've written it,
>
> I did not fully understand your response, cause it's about uwsgi
> implementation details, but it seems you did confirm the problem, didn't
> you?

yes, currently harakiri (and user_harakiri) works reliably only in
multiprocess mode.

> Do you have any estimates of when 2.1 would be out?

it did not started yet, so i assume it will happen not before may.

i am investigating if the thing can be applied to 2.0

>
> Thanks, Roberto!
>
>
> On Tue, Mar 11, 2014 at 11:26 AM, Roberto De Ioris <roberto at unbit.it>
> wrote:
>
>>
>> >
>> >> Roberto, could you please give a comment on this? Thanks a lot!
>> >
>> >
>> > oops, sorry, completely missed it, i will give a look tomorrow morning
>>
>>
>> first round:
>>
>> - start request for /index.html
>> - set harakiri to 5 seconds
>> - start request for /ws
>> - re-set harakiri to 5 seconds
>> - /index.html ends, harakiri set to 0 (race condition!!!)
>> - all continues until the next request is managed...
>>
>> second round
>>
>> - start request for /ws (as idnex.html is no more needed)
>> - set harakiri to 5 seconds
>> - harakiri is now honoured ...
>>
>>
>> Harakiri and multiple cores does not play well, but i am pretty
>> confident
>> we can find a solution in 2.1. Some idea ? (maybe it is enough to
>> maintain
>> the harakiri per-core instead of per-worker)
>>
>>
>> --
>> Roberto De Ioris
>> http://unbit.it
>> _______________________________________________
>> uWSGI mailing list
>> uWSGI at lists.unbit.it
>> http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
>>
> _______________________________________________
> uWSGI mailing list
> uWSGI at lists.unbit.it
> http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi
>


-- 
Roberto De Ioris
http://unbit.it


More information about the uWSGI mailing list