Managing NFS and KRPC Kernel Configurations in HP-UX 11i v2 (November 2013)

4
2 Managing Kernel Tunables Using Kctune
2.1 NFS Client Tunables
Table 2.1-1 lists the NFS client tunables. The last column specifies which NFS patch first introduced
the tunable.
Table 2.1-1 NFS Client Tunables
Kctune Tunable Name
Range
Default Value
Units
Patch ID
nfs_async_read_avoidance_enabled*
0 or 1
0
Boolean
PHNE_33982
nfs_clnt_avoid_write_copy*
0 or 1
0
Boolean
PHNE_34550
nfs_do_purge_rddir_cache*
0 or 1
0
Boolean
PHNE_33982
nfs_new_filelock_cancel*
0 or 1
0
Boolean
PHNE_38252
nfs_fine_grain_fs_lock*
0 t o 2
0
Boolean
PHNE_35118
nfs_new_lock_code*
0 or 1
0
Boolean
PHNE_33982
nfs_new_rnode_lock_code*
0 or 1
0
Boolean
PHNE_33982
nfs_nrnode
0 to MAXINT
0
Nrnodes
PHNE_33982
nfs_posix_pathconf_chown*
0 or 1
0
Boolean
PHNE_33982
nfs_wakeup_one*
0 to 2
0
Boolean
PHNE_35118
nfs2_do_symlink_cache
0 or 1
0
Boolean
PHNE_33982
nfs2_dynamic
0 or 1
1
Boolean
PHNE_33982
nfs2_shrinkreaddir
0 or 1
0
Boolean
PHNE_33982
nfs3_adaptive_cache_validate*
0 or 1
0
Boolean
PHNE_35960
nfs3_do_symlink_cache
0 or 1
1
Boolean
PHNE_33982
nfs3_dynamic
0 or 1
0
Boolean
PHNE_33982
nfs3_enhanced_rddir_cache *
0 or 1
0
Boolean
PHNE_43704
nfs3_hacc_size *
2 to MAXINT
32
Integer
PHNE_34550
nfs3_jukebox_delay
0 to MAXINT
1000
Hz
PHNE_33982
nfs3_max_transfer_size
1 to MAXINT
32768
Bytes
PHNE_33982
nfs3_max_transfer_size_clts
1 to 65536
32768
Bytes
PHNE_33982
nfs3_max_transfer_size_cots
1 to 65536
32768
Bytes
PHNE_33982
nfs3_new_acache *
0 or 1
0
Boolean
PHNE_34550
*This tunable is obsolete after the HP-UX 11i v2 release.
2.1.1 nfs_async_read_avoidance_enabled
Description
The nfs_async_read_avoidance_enabled tunable controls whether NFS clients are able to issue NFS
READ calls when all the biod daemons are busy. This improves NFS read performance on busy
clients.
When biods are running, reading applications block and wait for biods to retrieve data from the NFS
server. The biods place the data in the client's buffer cache for the application to access.
Simultaneously, writing processes are allowed to continuously post their data directly to the client's
buffer cache without blocking, provided buffer cache resources are available. Asynchronous write
requests are able to be queued to the buffer cache much faster than asynchronous read requests, thus
leading to starvation of reading processes because the writing processes monopolize the buffer cache
resources and biods. Enabling the nfs_async_read_avoidance_enabled tunable allows the NFS client