One more thing, those

SIGPIPE: writing to a closed pipe/socket/fd (probably the client
disconnected) on request / (ip 127.0.0.1) !!!

errors are printed at burst when ab finishes, so it's possible that uWSGI
make more backend requests than it receives.


2013/2/12 Łukasz Mierzwa <[email protected]>

> During 1.9 HTTP router testing I've noticed that request count reported by
> stats socket shows higher count than the number of requests I've made with
> ab. This happens only if I use high concurrency in ab (like -c 512), this
> is probably due to retries since some requests on my VM will fall under
> such conditions, if so than I guess it would be ok (although I prefer to
> have retries counter with the number of such requests) but the failcnt
> shows 0, so I'm not sure if those are really retries.
> Also if those were retries than backend node should show correct number of
> requests, but it show exactly the same number of requests as router.
> Can you help me interpret whats going on there? Full logs are at
> https://gist.github.com/prymitive/4761112
>
> --
> Łukasz Mierzwa
>



-- 
Łukasz Mierzwa
_______________________________________________
uWSGI mailing list
[email protected]
http://lists.unbit.it/cgi-bin/mailman/listinfo/uwsgi

Reply via email to