LOOKUPD

Section: Maintenance Commands (8)
Updated: July 21, 1992
Index Return to Main Contents

NAME

/usr/etc/lookupd - lookup daemon

SYNOPSIS

lookupd [ -i ] [ -f ] [ -m minutes ] [ -L filename ]

DESCRIPTION

The lookupd daemon is used in conjunction with NetInfo to speed up various system lookups, such as host entries, user names, and printer names. For performance reasons lookupd caches information about the location of NetInfo domains. It also caches various other information, including groups, user accounts, and mount entries. There are two different types of caches: the user entry cache (also called the pwent cache), and the other caches (hostname, mounts, logged-in user, root user, and printers). By default, the user entry cache is preloaded once, and lookupd then waits until the getpwent library routine is used before checking whether the cache contents have expired. (This is called lazy cache refresh.)

Other caches are maintained in a consistent fashion: prior to using the cached information, lookupd checks whether the source NetInfo database has changed. If not, the cached information is used; if so, that cache is refreshed and the new information used.

The logged-in user and the root user caches are flushed every 20 minutes, as well as when passwords are changed and when a user logs in or out.

The lookupd daemon supports the Network Information Service (NIS, formerly called Yellow Pages or YP) and BIND. By default, these features are turned off. NIS can be enabled using the HostManager application, by setting the NIS Domain Name. BIND is enabled by the existence of an /etc/resolv.conf file. The search policy when these are enabled is to try NetInfo first. If the information is located there, the result is returned. Otherwise, BIND is tried next (for host names and addresses), then NIS.

The lookupd daemon will restart itself in response to a SIGHUP signal, which can be sent from the shell via ``kill -HUP'' to lookupd's process id (pid). This is useful to tell lookupd that things have changed. For example, if the network gets turned on later, sending a SIGHUP to lookupd will get lookupd to recognize the network and look for information on it.

OPTIONS

-f: Force regularly scheduled updates of the pwent cache (by turning off lazy cache refresh); default is every 30 minutes.
-f -m X: The pwent cache is flushed every X minutes. Note: the value of X cannot be 0 (zero).
-m minutes: Set the lazy refresh interval for the pwent cache to the specified number of minutes. A value of zero turns off pwent caching altogether.
-i: Force an immediate pwent cache reload only if it is out of date, then exit. Incompatible with all other options.
-L filename: Asks that all lookup requests be logged to the specified file. Information logged includes the routine called, argument to the routine (an asterisk here denotes a cache hit), total number of calls to the routine, CPU time this call took (in microseconds), and total CPU time used by this routine (also in microseconds). Note that the lookupd log can become quite large over time.