On 08-jul-09, at 19:11, Leonardo Santagada wrote:
> On Jul 8, 2009, at 10:39 AM, Gunnar Wolf wrote:
>
>> Daniel L. Miller dijo [Tue, Jul 07, 2009 at 12:06:55PM -0700]:
>>> Is there a "rule of thumb" for determining the appropriate number of
>>> FastCGI processes?  In my case, I'm supporting a small LAN of less
>>> than
>>> 10 workstations, but I don't want any delays when they visit the
>>> different PHP virtual servers.  I would think just 1 or 2 would be
>>> sufficient - but is there something I'm missing?
>>
>> It is not _so_ important the amount of users, but the frequency of  
>> the
>> hits and the complexity of your program. Do you expect to have a
>> constant rate, say, more than one hit per five seconds? A execution
>> time of over one second? If neither of those is true, you could be
>> fine with a single process - of course, launching two (or even three)
>> to avoid collisions won't hurt too much.
>
> Isn't there a way to specify a range like in apache for the prefork
> worker? Minimal number of processes and a maximum would be great. Just
> in case some site gets slashdoted :)


That'd be a great feature, indeed.

However, that's something the PHP folks would have to implement in php- 
cgi. There is no much we can do (except start hacking PHP).

--
Octality
http://www.octality.com/

_______________________________________________
Cherokee mailing list
[email protected]
http://lists.octality.com/listinfo/cherokee

Reply via email to