On Wed, May 23, 2012 at 7:19 AM, Martin Buck <[email protected]> wrote: > Hi Ole, > > a patch would be great if it also includes changes to the man page > mentioning the buffer limit, the kernel limit, how to tune the latter and > the error message you get if it is exceeded. What would you suggest as the > new maximum block size?
If I read the code correctly it costs nothing to set the maximum block size to max of 64-bit integer. > The more fundamental problem is that SysV-style shared memory really doesn't > seem to be made for such large amounts of shared memory (which is probably > why the kernel has such low limits as well). The right way to proceed would > be to switch to more modern mechanisms like mmap(), but I guess that's a > larger amount of work. A patch for that would be most welcome. The 64-bit limit might not work on all systems, but I find it ridiculous that it is 'buffer' and not the system that sets the limit. It seems 'mbuffer -q' can be used instead, but it seems to be slower than 'buffer'. With -t it does memory mapped temporary files. /Ole -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
