Common Misconfigured HP-UX Resources (April 2006)

systems so the vxfsd daemon does not use too much CPU. The following example uses kctune to

change vxfs_ifree_timelag without a reboot:

# kctune vxfs_ifree_timelag=-1

: :

Tunable Value Expression Changes

vxfs_ifree_timelag (before) 0 Default Immed

(now) -1 -1

Summary

Deciding whether or not to tune the JFS inode cache depends on how you plan to use your system.

Memory is a finite resource, and a system manager needs to decide how much of the system’s

memory needs to be spent on a specific resource.

By default, the JFS inode cache is configured to be very large. You should understand the advantages

and disadvantages of using the default JFS inode cache sizes and behavior. Consider the type of

access to the files on the system and the memory cost associated with the JFS inode cache when

deciding whether or not to tune the JFS inode cache.

For most customers, tuning the JFS inode cache down would save memory and potentially enhance

performance by leaving more memory for other operations. With the introduction of vxfsstat(1M) on

JFS 3.5, you can determine how many inodes are actually in use at a given time to judge how large

your inode cache should be. Most systems do not need more than 20,000 inodes, so reducing the

size of the JFS inode cache can reduce system memory utilization. However, be careful of reducing

the size of the cache on JFS 4.1 or later by checking the number of inode free lists.

Since the JFS inode cache often expands to the maximum amount, using a fixed-sized inode cache

will help the kernel memory allocator from managing large per-CPU freelists. Therefore, tuning

vx_noifree or vxfs_ifree_timelag can also reduce overall system memory utilization as well

as reducing the amount of CPU utilization used by the vxfsd daemon.

However, if the primary use is for the system is as a fileserver, where random file lookups are being

performed constantly or sequential lookups are done and the working set is less than the size of the

inode cache, then using the default JFS inode cache sizes is probably best. Application performance

could be degraded if the application relies on having a larger working set of inodes in the cache.