©
Ce document utilise Manuel du site Web PHP chinois Libérer
(PECL gearman >= 0.5.0)
GearmanClient::addTaskBackground — Add a background task to be run in parallel
$function_name
, string $workload
[, mixed &$context
[, string $unique
]] )Adds a background task to be run in parallel with other tasks. Call this method for all the tasks to be run in parallel, then call GearmanClient::runTasks() to perform the work.
function_name
由程序自动执行的已注册函数。
workload
被处理的序列化数据。
context
与任务关联的应用程序上下文。
unique
用于标识特定任务的唯一性 ID。
A GearmanTask object or FALSE
if the task could not be added.
Example #1 Two tasks, one background and one not
This example illustrates the difference between running a background task and a normal task. The client adds two tasks to execute the same function, but one is added with addTaskBackground() . A callback is set so that progress of the job can be tracked. A simple worker with an artificial delay reports on the job progress and the client picks this up through the callback. Two workers are run for this example. Note that the background task does not show in the client output.
<?php
# The client script
# create our gearman client
$gmc = new GearmanClient ();
# add the default job server
$gmc -> addServer ();
# set a couple of callbacks so we can track progress
$gmc -> setCompleteCallback ( "reverse_complete" );
$gmc -> setStatusCallback ( "reverse_status" );
# add a task for the "reverse" function
$task = $gmc -> addTask ( "reverse" , "Hello World!" , null , "1" );
# add another task, but this one to run in the background
$task = $gmc -> addTaskBackground ( "reverse" , "!dlroW olleH" , null , "2" );
if (! $gmc -> runTasks ())
{
echo "ERROR " . $gmc -> error () . "\n" ;
exit;
}
echo "DONE\n" ;
function reverse_status ( $task )
{
echo "STATUS: " . $task -> unique () . ", " . $task -> jobHandle () . " - " . $task -> taskNumerator () .
"/" . $task -> taskDenominator () . "\n" ;
}
function reverse_complete ( $task )
{
echo "COMPLETE: " . $task -> unique () . ", " . $task -> data () . "\n" ;
}
?>
<?php
# The worker script
echo "Starting\n" ;
# Create our worker object.
$gmworker = new GearmanWorker ();
# Add default server (localhost).
$gmworker -> addServer ();
# Register function "reverse" with the server.
$gmworker -> addFunction ( "reverse" , "reverse_fn" );
print "Waiting for job...\n" ;
while( $gmworker -> work ())
{
if ( $gmworker -> returnCode () != GEARMAN_SUCCESS )
{
echo "return_code: " . $gmworker -> returnCode () . "\n" ;
break;
}
}
function reverse_fn ( $job )
{
echo "Received job: " . $job -> handle () . "\n" ;
$workload = $job -> workload ();
$workload_size = $job -> workloadSize ();
echo "Workload: $workload ( $workload_size )\n" ;
# This status loop is not needed, just showing how it works
for ( $x = 0 ; $x < $workload_size ; $x ++)
{
echo "Sending status: " . ( $x + 1 ) . "/ $workload_size complete\n" ;
$job -> sendStatus ( $x + 1 , $workload_size );
$job -> sendData ( substr ( $workload , $x , 1 ));
sleep ( 1 );
}
$result = strrev ( $workload );
echo "Result: $result \n" ;
# Return what we want to send back to the client.
return $result ;
}
?>
Worker output for two workers running:
Received job: H:foo.local:65 Workload: !dlroW olleH (12) 1/12 complete Received job: H:foo.local:66 Workload: Hello World! (12) Sending status: 1/12 complete Sending status: 2/12 complete Sending status: 2/12 complete Sending status: 3/12 complete Sending status: 3/12 complete Sending status: 4/12 complete Sending status: 4/12 complete Sending status: 5/12 complete Sending status: 5/12 complete Sending status: 6/12 complete Sending status: 6/12 complete Sending status: 7/12 complete Sending status: 7/12 complete Sending status: 8/12 complete Sending status: 8/12 complete Sending status: 9/12 complete Sending status: 9/12 complete Sending status: 10/12 complete Sending status: 10/12 complete Sending status: 11/12 complete Sending status: 11/12 complete Sending status: 12/12 complete Sending status: 12/12 complete Result: !dlroW olleH Result: Hello World!
Client output:
STATUS: 1, H:foo.local:66 - 1/12 STATUS: 1, H:foo.local:66 - 2/12 STATUS: 1, H:foo.local:66 - 3/12 STATUS: 1, H:foo.local:66 - 4/12 STATUS: 1, H:foo.local:66 - 5/12 STATUS: 1, H:foo.local:66 - 6/12 STATUS: 1, H:foo.local:66 - 7/12 STATUS: 1, H:foo.local:66 - 8/12 STATUS: 1, H:foo.local:66 - 9/12 STATUS: 1, H:foo.local:66 - 10/12 STATUS: 1, H:foo.local:66 - 11/12 STATUS: 1, H:foo.local:66 - 12/12 COMPLETE: 1, !dlroW olleH DONE
[#1] raitech at gmail dot com [2015-11-12 10:47:57]
This method seems only useful when you doesn't need to know something about the worker, when you can let runTasks() finishes its code and destroy the GearmanClient object without a problem.
If you need to get data from worker via callbacks, forget it after runTasks() finishes.
You have to block your execution in the GearmanClient "main loop", so it can call the callbacks. It seems that runTasks() is that "main loop".
[#2] Anonymous [2015-08-31 20:02:09]
It is unlikely this example works quite as advertised.
The foreground job will block, however the background job should not.. (and if it does that's not documented Gearman behaviour far as I know and would not make sense for a Background job).
So, if the foreground job completes, then we would expect runTasks() to return at that moment, regardless of the status of the background job(s). With nothing else to do, the php script (the client) in this example would exit at that point.
To fully-utilize background jobs, it's reasonable to assume that we still wish to know their status. To do that, you need a polling loop that "checks up on them" and waits until their completion.
That eliminates the point of background jobs of course - because the client would still be Blocking (in a polling loop) more or less occupied, and unable to exit/finish.
So, in practice, background jobs mean you are decoupled from the client upon execution (because the main point is to free up the client, otherwise just use foreground execution), which means the client is likely going to exit, meaning the jobs themselves should be reporting their status and final result independently, not leaving it up to the client.
It's a significantly different setup compared to foreground jobs. In fact this example is kind of silly to even mix the two.
Nobody will ever see this post, because apparently nobody in the world has ever commented on the php gearman client, but it's a good post never the less.