large number of simultaneous open files

[aryan_]

Hi All,

I know that there is an OS limitation on the number of files that can be kept open for writing simultaneously.

However, I want to somehow overcome this limitation.

Let's say, if I have to process X number of files simultaneously at a certain time, where X is greater than the number of simultaneously open files that OS can handle, it seems to be the only solution to close the oldest files which were opened before any others first, and open the new ones instead.

Maybe a circular buffer or something similar is needed for this.

What is the best way to handle such a scenario in Linux using C++ ?

What Linux system function gives me the number of files that I can keep open at a certain time ?

Thanks.

[drl]

Hi.

I don't use this very often, so I needed to search for keywords in man page titles. It's a configuration value, so I look for config, and then filter those results for system:

% man -k config | grep system
getconf (1)          - Query system configuration variables
paperconfig (8)      - configure the system default paper size
system-config-printer (1) - configure a CUPS server
system-config-printer-applet (1) - print job manager

and getconf looks like the command to use, but man getconf does not tell me directly what variable to look for. However, the reference at the bottom of man getconf suggests I look at man sysconf, in which I find:

       OPEN_MAX - _SC_OPEN_MAX
              The  maximum number of files that a process can have open at any
              time.  Must not be less than _POSIX_OPEN_MAX (20).

which, when used, produces on Debian 5 and CentOS 5.4:

% getconf OPEN_MAX
1024

After reading both man pages, it might be reasonable to test that value with a short program to be sure it's what you want.

Best wishes ... cheers, drl