3.7.0 HP StorageWorks HP Scalable NAS File Serving Software administration guide - HP Scalable NAS 3.7 for Linux (AG513-96002, October 2009)
system’s physical memory. For example, for a 16GB system, this would mean
setting the parameter to 800000. HP has seen some cases where a slightly
higher value (6 or 7% of physical memory) further improves replication perform-
ance.
• vm.Vfs_cache_pressure controls the rate at which filesystem inode structures are
freed from the system and should be set to a value of 10000 when replicating
large sets of files. This will ensure that the system cache does not get overrun with
filesystem inode structures when replicating.
In the HP Scalable NAS software-only product, the SizingActions script (found
in /etc/opt/hpcfs) sets these parameters to values that are appropriate for a
system not running replication. You can edit the SizingActions file to change
these parameters as appropriate; use your favorite text editor to edit the file and save
it back in the same location. (Elsewhere in this guide, HP recommends against editing
the SizingActions file. In this instance, editing it is the appropriate solution to
setting the parameters.)
If, as is the case with some configurations, you do not run the SizingActions
script at startup (for example, you are running Oracle or some other major application
that has its own tuning recommendations), you will need to set the parameters in the
/etc/sysctl.conf file so that they are reset at system startup.
ssh configuration
The replication feature uses ssh to communicate with the destination cluster. To use
ssh, a matching key pair must be in place on the source and destination clusters.
HP Scalable NAS includes a default key pair that should be installed on each cluster.
(If desired, you can create a custom key pair and then install that key pair on the
source and destination clusters. See Create a custom ssh key pair, page 269.)
The rplkeys command is used to install the key pair. The command also tunes the
sshd_config file on each cluster to support 200 simultaneous ssh connection
starts (the default is 10). The command adds a comment before the changed line in
the file indicating when the value was changed.
The replication processes are run as root, but the private key used is not in the default
location (~root/.ssh) on the source cluster. Instead, it is in /_adminfs/
replication/authentication/replication_ssh_key. The public key is
published to the destination cluster node in ~/.ssh/authorized_keys with
'replication' in the comment section at the end of the key.
Configure and manage replication254