#
# Asynchronous client-to-server (DEALER to ROUTER)
#
if {[llength $argv] == 0} {
set argv [list driver 3 5]
} elseif {[llength $argv] != 3} {
puts "Usage: asyncsrv.tcl ?<driver|client|server|worker> <number_of_clients> <number_of_workers>?"
exit 1
}
set tclsh [info nameofexecutable]
lassign $argv what NBR_CLIENTS NBR_WORKERS
expr {srand([pid])}
switch -exact -- $what {
client {
# This is our client task
# It connects to the server, and then sends a request once per second
# It collects responses as they arrive, and it prints them out. We will
# run several client tasks in parallel, each with a different random ID.
package require zmq
zmq context context
zmq socket client context DEALER
# Set random identity to make tracing easier
set identity [format "%04X-%04X" [expr {int(rand()*0x10000)}] [expr {int(rand()*0x10000)}]]
client setsockopt IDENTITY $identity
client connect "tcp://localhost:5570"
proc receive {} {
global identity
puts "Client $identity received [client recv]"
}
proc request {} {
global request_nbr identity
incr request_nbr
puts "Client $identity sent request \#$request_nbr"
client send "request \#$request_nbr"
after 1000 "request"
}
# Process responses
client readable receive
# Send a request every second
set request_nbr 0
after 1000 request
vwait forever
client close
context term
}
worker {
# This is our worker task
# Accept a request and reply with the same text a random number of
# times, with random delays between replies.
package require zmq
zmq context context
zmq socket worker context DEALER
worker connect "ipc://backend"
while {1} {
# The DEALER socket gives us the address envelope and message
set address [worker recv]
set content [worker recv]
puts "worker received $content from $address"
# Send 0..4 replies back
set replies [expr {int(rand()*5)}]
for {set reply 0} {$reply < $replies} {incr reply} {
# Sleep for some fraction of a second
after [expr {int(rand()*1000) + 1}]
puts "worker send $content to $address"
worker sendmore $address
worker send $content
}
}
}
server {
# This is our server task It uses the multithreaded server model to deal
# requests out to a pool of workers and route replies back to clients. One
# worker can handle one request at a time but one client can talk to multiple
# workers at once.
package require zmq
zmq context context
# Frontend socket talks to clients over TCP
zmq socket frontend context ROUTER
frontend bind "tcp://*:5570"
# Backend socket talks to workers over inproc
zmq socket backend context DEALER
backend bind "ipc://backend"
# Launch pool of worker threads, precise number is not critical
for {set thread_nbr 0} {$thread_nbr < $NBR_WORKERS} {incr thread_nbr} {
exec $tclsh asyncsrv.tcl worker $NBR_CLIENTS $NBR_WORKERS > worker$thread_nbr.log 2>@1 &
}
# Connect backend to frontend via a queue device
# We could do this:
# zmq_device (ZMQ_QUEUE, frontend, backend);
# But doing it ourselves means we can debug this more easily
proc do_frontend {} {
set address [frontend recv]
set data [frontend recv]
backend sendmore $address
backend send $data
}
proc do_backend {} {
set address [backend recv]
set data [backend recv]
frontend sendmore $address
frontend send $data
}
backend readable do_backend
frontend readable do_frontend
vwait forever
frontend close
backend close
context term
}
driver {
puts "Start server, output redirected to server.log"
exec $tclsh asyncsrv.tcl server $NBR_CLIENTS $NBR_WORKERS > server.log 2>@1 &
after 1000
for {set i 0} {$i < $NBR_CLIENTS} {incr i} {
puts "Start client $i, output redirect to client$i.log"
exec $tclsh asyncsrv.tcl client $NBR_CLIENTS $NBR_WORKERS > client$i.log 2>@1 &
}
}
}