cnid_metad — start cnid_dbd daemons on request
cnid_metad waits for requests from afpd to start up instances of the cnid_dbd daemon. It keeps track of the status of a
cnid_dbd instance once started and will
restart it if necessary. cnid_metad is normally started
at boot time from /etc/rc or equivalent and runs
until shutdown. afpd needs to be configured
with the -cnidserver option in afpd.conf in order to access cnid_metad. It is possible to run more than one
instance of cnid_metad on the same machine if different
values for the interface and/or port are specified with the
-h and -p options.
-l <loglevel> [-f
<filename>]Specify that any message of a loglevel up to the given
loglevel should be logged to the given file. If the
filename is ommited the loglevel applies to messages passed to
syslog. Default is logs to syslog with a default logging setup of
"log_note".
Note: cnid_metad passes this to cnid_dbd which will in turn use this for its logging setup.
loglevels: LOG_SEVERE, LOG_ERROR, LOG_WARN, LOG_NOTE, LOG_INFO, LOG_DEBUG, LOG_DEBUG6, LOG_DEBUG7, LOG_DEBUG8, LOG_DEBUG9, LOG_MAXDEBUG
-dcnid_metad will remain in the foreground and will also leave the standard input, standard output and standard error file descriptors open. Useful for debugging.
-h hostnameUse hostname as the network interface for requests as opposed to the default localhost. The network address may be specified either in dotted-decimal format for IPv4 or in hexadecimal format for IPv6.
-p portUse port as the port number for reqests. Default is 4700.
-u userSwitch to the userid of user before serving requests. This userid will be inherited by all cnid_dbd daemon processes started.
-u groupSwitch to the groupid of group before serving requests. This groupid will be inherited by all cnid_dbd daemon processes started. Both user and group must be specified as strings.
-s cnid_dbd
pathnameUse cnid_dbd pathname as the pathname of the executeable of the cnid_dbd daemon. The default is cnid_dbd.
The number of cnid_dbd subprocecesses is currently limited to 512.
cnid_metad does not block or catch any signals apart from SIGPIPE. It will therefore exit on most signals received. This will also cause all instances of cnid_dbd's started by that cnid_metad to exit gracefully. Since state about and IPC access to the subprocesses is only maintained in memory by cnid_metad this is desired behaviour. As soon as cnid_metad is restarted afpd processes will transparently reconnect.