If my vote counts +1 since we saw how dramatic performance goes down.
As far as I can see from the Solaris documentation (and the system
behaviour) fdatasync really writes to the physical disk. It would be
enough to flush file buffers to the file system (which will be cached in
modern operating systems and the data will be visible to other processes
looking at the file although it is not already physically on the disk). I
think the file system is the right layer to use and the fflush call is
already in the code right before the fdatasync.
Rainer Jung
At 15:34 20.04.01 , you wrote:
>I'm strongly to remove the fdatasync in mod_jk since :
>
>- it slow down too much Apache when running in DEBUG/TRACE mode
>- not very portable (not PORTABLE/FREEBSD/WIN32)
>
>-
>Henri Gomez ___[_]____
>EMAIL : [EMAIL PROTECTED] (. .)
>PGP KEY : 697ECEDD ...oOOo..(_)..oOOo...
>PGP Fingerprint : 9DF8 1EA8 ED53 2F39 DC9B 904A 364F 80E6
>
>
>
> >-----Original Message-----
> >From: Arkadiusz Rychlinski [mailto:[EMAIL PROTECTED]]
> >Sent: Friday, April 20, 2001 8:53 AM
> >To: [EMAIL PROTECTED]
> >Subject: I made some changes in Tomcat's sources....
> >
> >
> >I made some changes in (maybe not exacly) Tomcat's sources.
> >How can I make
> >it public?
> >
> >I've added support for encoding different than ISO-8859-1 in
> >form data from
> >GET/POST request and processing multipart forms and file upload.
> >
> >What should I do to present these changes to tomcat team?
> >
> >
> >
> >Arek
> >
> >
> >
> >
