Managing NFS and KRPC Kernel Configurations in HP-UX 11i v3 (762807-001, March 2014)

2.1.45 nfs_ufc_threshold_percentage
Description
The nfs_ufc_threshold_percentage tunable controls the amount of UFC file cache that can be
consumed by the NFS client at any given point of time. This tunable will be effective only when the
nfs_enable_ufc_threshold tunable is enabled.
Tested Values
Default: 50% (NFS client’s UFC usage limited to 50% of filecache_max at any given point)
Min: 40%
Max: 70% (Maximum percentage of filecache_max that NFS client can use at any given point of
time)
Note: This tunable is implemented only for NFS v3 and v4 mounts.
Restrictions on Changing The nfs_ufc_threshold_percentage tunable is static. All NFS mount points
must be unmounted before making any change to this tunable. System reboot is not required to
activate a change made to this tunable.
Modifying the Value
1.Based on the UFC file cache settings and workload, users can change this tunable to enable the
NFS client to use a limited UFC file cache across all NFS v3 and v4 mount points.
2.1.46 nfs_enable_enhanced_commit
Description
The nfs_enable_enhanced_commit tunable is used to improve the way NFS client sends the commit
calls over the wire to the NFS server. When this tunable is enabled, the number of over the wire
commit calls are reduced significantly. In addition to that, a separate set of kernel threads are
created dedicated only to make NFS commit calls. Due to this, having this tunable enabled might
even improve the overall write performance.
By default, each NFS mount has a default set of kernel threads that perform asynchronous I/O. This
I/O includes sending of NFS write, read, readdirplus and commit calls over the wire. On enabling
this tunable, each NFS mount contains a separate and dedicated set of kernel threads exclusive for
sending commit calls in addition to the default set of kernel threads meant for the other asynchronous
I/O. By having this separate set of kernel threads dedicated only for commit processing, the default
set of kernel threads can have enough bandwidth to perform other asynchronous I/O activities like
NFS writes. The number of kernel threads dedicated for sending commit calls depends on the
nfs3_max_commit_threads tunable.
Tested Values
Default: 0 (Tunable is disabled)
Min: 0
Max: 1 (Tunable is enabled)
Restrictions on Changing
The nfs_enable_enhanced_commit tunable is dynamic. System reboot is not required to activate
changes made to this tunable. However, the enhanced commit model is set per filesystem at mount
time. The system administrator must unmount and re-mount each filesystem after changing this tunable.
At this time, only NFSv3 mount points are affected by changing the value of this tunable.