rwhod -- system status server


/etc/rwhod [ -m ] [ -M ttl ]


rwhod is the server that maintains the database used by the rwho(TC) and ruptime(TC) programs. Its operation is predicated on the ability to broadcast messages on a network.

rwhod operates as both a producer and consumer of status information. As a producer of information, it periodically queries the state of the system and constructs status messages which are broadcast on a network. As a consumer of information, it listens for other rwhod servers' status messages, validating them, then recording them in a collection of files located in the directory /usr/spool/rwho.

The server transmits and receives messages at the port indicated in the ``rwho'' service specification; see services(SFF). The messages sent and received are of the form:

   struct  outmp {
       char    out_line[8];     /* tty name */
       char    out_name[8];     /* user id */
       long    out_time;        /* time on */

struct whod { char wd_vers; char wd_type; char wd_fill[2]; int wd_sendtime; int wd_recvtime; char wd_hostname[32]; int wd_loadav[3]; int wd_boottime; struct whoent { struct outmp we_utmp; int we_idle; } wd_we[1024 / sizeof (struct whoent)]; };

All fields are converted to network byte order prior to transmission. The load averages are as calculated by the sar(ADM) program, and represent load averages over the 5, 10, and 15 minute intervals prior to a server's transmission; they are multiplied by 100 for representation in an integer. The host name included is that returned by the gethostname(SLIB) system call, with any trailing domain name omitted. The array at the end of the message contains information about the users logged in to the sending machine. This information includes the contents of the utmp(F) entry for each non-idle terminal line and a value indicating the time in seconds since a character was last received on the terminal line.

Messages received by the rwho server are discarded unless they originated at an rwho server's port. In addition, if the host's name, as specified in the message, contains any unprintable ASCII characters, the message is discarded. Valid messages received by rwhod are placed in files named whod.hostname in the directory /usr/spool/rwho. These files contain only the most recent message, in the format described above.

Status messages are generated approximately once every 5 minutes. rwhod performs an nlist(S) on UNIX every 30 minutes to guard against the possibility that this file is not the system image currently operating.

rwhod can be configured to use IP multicasting if desired. The -m flag causes rwhod to send its updates to the IP multicast address Datagrams will be sent (with a time-to-live of one) to all interfaces marked as supporting multicasting. If the -M option is used, datagrams will only be sent out a single interface, using a time-to-live of ttl. The interface used will be determined from the routing table.

When rwhod is using multicasting, older systems will be unable to hear updates sent to multicast addresses. However, rwhod will still receive updates that are sent using broadcasting.


The rwhod daemon is commented out of the /etc/tcp startup script, as it can degrade system performance. If you want to use rwhod, uncomment the following three lines:
#	if [ -x /etc/rwhod -a -d /usr/spool/rwho ]; then
#		rwhod ; echo "rwhod \c"
#	fi

This service takes up progressively more network bandwidth as the number of hosts on the local net increases. For large networks, the cost becomes prohibitive.

rwhod should relay status information between networks. People often interpret the server dying as a machine going down.

Some mechanism for cleaning dead machine data out of the spool directory is needed.

See also

gethostname(SLIB), nlist(S), ruptime(TC), rwho(TC), utmp(F), w(C)

Standards conformance

rwhod is conformant with X/Open Portability Guide, Issue 4, 1992.
© 2003 Caldera International, Inc. All rights reserved.
SCO OpenServer Release 5.0.7 -- 11 February 2003