- Kernel 2.4.x Architectural Changes (Filesystems)
- The Shared Memory Filesystem (shm)
- Conclusion
The Shared Memory Filesystem (shm)
One thing you probably haven't noticedbecause unless you have some exotic requirements, this will all be quite transparent to youis the shared memory filesystem.
This filesystem was introduced and is included by default in the new kernel. You can't turn it off, and you need do nothing special for it. It's there. That said, some folks will need to modify, if only slightly, some of the parameters of the shared memory filesystem. Usually this requirement is to increase the size of the filesystem for use with large databases or other programs that make extensive use of shared memory.
If this is your case, create a mount point for the shared memory filesystem, such as /var/shm. This makes the shared memory filesystem available for manipulation. This filesystem exists, but only in memory. You can't change it, though, except when you mount it. During mount, you can modify the nr_blocks (number of blocks) or the nr_inodes (number of inodes) available to the shared memory filesystem. If you want to see the default blocks, mount the filesystem and use df, which will show you the number of 1K blocks. If the Use% gets large, you may need to specify more blocks.