rpc.lockd, lockd - Network lock daemon
/usr/sbin/rpc.lockd [-b hostname] [-c] [-d debug] [-g graceperiod]
[-h hashsize] [-s] [-t timeout]
The default behavior will create and bind a socket for
each protocol per interface on the system. When the -b
switch is specified, one socket per protocol will be created
and bind to the passed IP address. Available only on
TruCluster Server systems. This option starts the clusterwide
lock daemon, which helps provide highly available NFS
service. Do not use -c directly. Highly available NFS
service is configured by default and typically does not
require intervention. If you do need to start the clusterwide
lock daemon, use the CAA command, caa_start cluster_lockd.
For more information, see the TruCluster Server
Administration manual. Internal Use Only. Use
this option only under the direction of technical
support personnel. Causes the rpc.lockd daemon to
use the variable graceperiod (in seconds) as the
grace period duration instead of the default value
of 15 seconds. On TruCluster Server systems the
default value is 30 seconds. Internal Use Only.
Causes the rpc.lockd daemon to use the variable
timeout (in seconds) as the interval instead of the
default value of 5 seconds to retransmit a lock
request to the remote server.
The rpc.lockd daemon processes lock requests that are
either sent locally by the kernel or remotely by another
lock daemon. The NFS locking service makes this advisory
locking support possible by using the fcntl system call
and the lockf subroutine. The rpc.lockd daemon forwards
lock requests for remote data to the server site's lock
daemon. The rpc.lockd daemon then requests the status monitor
daemon, rpc.statd, for monitor service. The reply to
the lock request is not sent to the kernel until the status
daemon and the server site's lock daemon have replied.
If either the status monitor or server site's lock daemon
is unavailable, the reply to a lock request for remote
data is delayed until all daemons become available.
When a server recovers, it waits for a grace period for
all client site lock daemons to submit reclaim requests.
Client site lock daemons are notified by rpc.statd of the
server recovery and promptly resubmit previously granted
lock requests. If a client site's lock daemon fails to
secure previously granted locks at the server site, it
sends the signal SIGLOST to all the processes that were
previously holding locks and cannot reclaim them.
Functions: fcntl(2), signal(2), lockf(3)
[ Back ]