On Wednesday April 25 2007 6:14 am, Varj=C3=BC Tam=C3=A1s wrote:
> Hello everyone!
>
> I have written a perl script which opens a file for reading and writing.
> I tested what happens if the script is run in multiple instances by the
> webserver. I put a sleep command between seeking writing position in
> the file and actual writing. I filled the web form in two different
> tabs of my browser and sent it.
>
> My impression was that the following happens:
> - first instance opens the file,
> - first instance seeks the writing position,
> - first instance goes sleeping,
> - second instance opens the same file,
> - second instance seeks the writing position (different from the above
> one),
> - second instance goes sleeping,
> - first instance wakes up,
> - first instance writes the file,
> - first instance closes the file,
> - second instance wakes up,
> - second instance writes the file,
> - second instance closes the file.
>
> After that procedure both changes has appeared in the file.
>
> Does it mean, that on this system the multiple instance writing of a
> file is handled well and I do not have to worry about locking the file
> as long as the different instances write different positions in the
> file? Even the system administrator had the tip, that only the last
> change will appear.
>
> If I want to be sure that multiple instance writing will work, is append
> a better choice? I have read that on the archive of this list.
>
> Thanks
you should use flock to lock the file , so one process can work on it at
a=
=20
time , if you don't eventually a race way condition will occur and
you=20
will see a lock up happen , or if your luck one process will
overwrite=
=20
the other and you will just loose data.. either way it's just smart=20
programming to lock a file your working on especially if it's
happening=20
under CGI conditions.
Greg
|
