Il giorno 12/set/2012, alle ore 09:00, Łukasz Mierzwa <[email protected]> ha 
scritto:

> 2012/9/12 Roberto De Ioris <[email protected]>:
>> Your "update resolution", is too near the "master frequency".
>> 
>> As soon as the master slowdown a bit (like when calling fork()), the master 
>> clock will lose of a bunch of milliseconds (or even a whole second on a 
>> loaded system).
>> 
>> Probably setting the carbon update frequency to 2 seconds will fix the 
>> problem, but a better solution would be allowing milliseconds cycles in the 
>> master. Something to look at
>> for 1.4 (expecially now that we have high-resolution/highly-reliable clock 
>> plugins)
> 
> I'm not sure if I understand, master cycle frequency is 1 second,
> carbon update freq is 60 seconds (default).
> Master cycle takes some time since it may have a lot of work to
> handle, but master call carbon in every cycle and carbon will check if
> it's 60 seconds or more since last update, so I don't see any way to
> loose a carbon update cycle here. At worst if master had a lot of work
> in that cycle, than it would delay carbon update, but still timestamp
> check would work.

Ops, sorry i was sure you set update frequency to 1 second in the carbon plugin.

> 
> Machine I'm using to collect carbon stats is a VM with slow I/O, it's
> possible that I'm hitting connection timeout since with current code
> carbon plugin will not print any error message and probably set
> last_update=now() so at next cycle stats won't be resend.
> I'll try to fix that later today.
> 
> 

This is probably the reason, as uwsgi_connect() do no print error messages, but 
simply fails with -1

--
Roberto De Ioris
http://unbit.it
JID: [email protected]

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

Reply via email to