9.0;sf_helper (store_forward_helper), revision 9.0, 88/02/02
SF_HELPER (STORE_FORWARD_HELPER) -- Store and Forward Server.
usage: /sys/sf/sf_helper [queue_name...] [-local] [-all_local] [-forever]
[-delay n] [-status]
FORMAT
/sys/sf/sf_helper [queue_name...] [options]
sf_helper is the Store and Forward ipc server. Some Domain optional software
products use the Store and Forward ipc system for reliable communication
between nodes, even when there may be temporary partitioning of the network.
When a message's route is temporarily unavailable, the message can be stored
in a "local queue." Later, the message must be automatically forwarded, until
it's route is completed. sf_helper is the server which performs the automatic
forwarding.
sf_helper works by polling the queues for which it is responsible to see if
any messages have arrived which need to be forwarded. The -forever option may
be used to tell sf_helper to poll periodically, with a default cycle time of
15 minutes between inquiries. You may alter the cycle time using the -delay
option.
An sf_helper running on one node may forward messages from all nodes' local
queues, or for some subset of the nodes, or for just for the one node itself.
A typical configuration is to have a few sf_helpers running on the network,
trying to forward all messages on any of the nodes. These few sf_helpers use a
large percentage of their nodes' computing resources when they are executing,
but most of the time they are dormant. At the other extreme, every node may
run the sf_helper, but only try to forward messages stored on the node itself.
In this configuration, each node pays a small price in computing resources for
the forwarding services. Determining which configuration to use is left to the
system administrator -- the important thing is to be sure that all nodes'
local queues are being watched by at least one sf_helper.
sf_helper is typically invoked as a background server process via the DM
command cps. If a node is to run sf_helper all of the time, you may find it
convenient to add this cps command to the `node_data/startup file so that
sf_helper starts automatically whenever the node is rebooted.
ARGUMENTS
queue_name
(optional) Specify the pathname of the queue to be polled. Multiple
pathnames are permitted; separate them with blanks.
Default if omitted: must use -local or -all_local (below).
OPTIONS
-local Service the local queue of the local node. Equivalent to
specifying a queue pathname of '/sys/sf/local_q' as an
argument.
-all_local Service all local queues, i.e., every '/sys/sf/local_q'
currently on the network.
-forever Cause sf_helper to poll periodically, repeating forever
until its process is stopped. Without this option,
sf_helper will make one pass and then terminate.
-delay n Set the polling cycle time to 'n' minutes. If this option
is omitted and -forever is specified, the default cycle
time is 15 minutes.
-status Cause sf_helper only to display the status of the specified
queues, but not forward any messages.
EXAMPLES
1. $ /sys/sf/sf_helper //node1/sys/sf/local_q //node2/sys/sf/local_q -forever
Poll the queues on node1 and node2 every fifteen minutes until the process
stops.
2. $ /sys/sf/sf_helper -local -status
Display the status of the local node's local queue. Do not forward any
messages, and do not repeat.
3. Command: cps /sys/sf/sf_helper -forever -all_local -delay 60
When typed to the DM command window, this command starts sf_helper in a
background process, periodically polling all local queues on the network
(once every 60 minutes). If included in the node startup file
`node_data/startup, this command will restart the server whenever the
node is rebooted.
RELATED TOPICS
More information is available. Type:
- HELP DSEE SERVERS
More information is available about using sf_helper with the optional
software packages which use the Store and Forward system. See this HELP file
if your system has the DOMAIN SOFTWARE ENGINEERING ENVIRONMENT (DSEE)
software installed.