John McEntee wrote: > Is the open files going to cause CIFS a problem, 256 seems a bit low, could > easily hit that limit if it is shared amongst all the users.
"man ulimit" and "man getrlimit" for details. That's a limit on the number of file descriptors open in any one user process. It has nothing to do with the kernel, and shouldn't have anything to do with CIFS. The default is 256 in order to keep binary compatibility with fileno(3C). That library "function" was once (very long ago) defined as a macro that read a uchar_t directly from the FILE * pointer. In the modern implementation, it is a function, and allows a huge number of open files, but if you were running ancient SunOS binaries, you would want to be able to avoid problems. In any event, 256 is darned large for most reasonable applications, where pfiles will show a few dozen open descriptors. You can bump it up if you like; the normal maximum is 64K. I would recommend bumping it up *only* when launching programs that really need to have a large number of descriptors open at once, such as single-process web servers. Otherwise, a malfunctioning program can really bog things down. (I'd expect that most such programs already use setrlimit(2) to do something reasonable here ...) -- James Carlson 42.703N 71.076W <carls...@workingcode.com> _______________________________________________ OpenIndiana-discuss mailing list OpenIndiana-discuss@openindiana.org http://openindiana.org/mailman/listinfo/openindiana-discuss