MongoDB聚合在PHP中不起作用（或非常慢），并且在Shell中完美地工作？

Question

I'm trying to use the aggregate method on my collection (containing more than 20M documents).

I first tried it in the Windows shell :

db.data.aggregate([
{$match: {firstname: "Roger"}},
{$group:{"_id":"$id_car",count:{$sum: 1}}},
{$sort: {count: -1}},
{$limit: 50}])

And it works perfectly, returning the results after a few seconds.

When I "translate" it in PHP :

$data = $db->data;
$ops = array(
    array(
         '$match' => array(
             'firstname' => 'Roger'
         )
    ),
    array(
        '$group' => array(
            '_id' => '$id_car',
            'count' => array(
                '$sum' => 1     
            )       
        )
    ),
    array(
        '$sort' => array(
            'count' => -1
        )
    ),
    array(
        '$limit' => 4       
    )
);
$res = $data->aggregate($ops);

I get a timeout PHP Fatal error :

Uncaught exception 'MongoCursorTimeoutException' with message 'localhost:27017: cursor timed out (timeout: 30000, time left: 30:0, status: 0)'

I don't know if I've made a mistake in my PHP code, or if aggregate is supposed to be much slower in PHP than in shell ?

Also, I have added an index on "firstname" field to make the query go faster.

By the way, is there any way to set the timeout to infinity for this kind of call ?

Thanks a lot for your help !

Joe

Answer 1

I don't really know about your issue (PHP being slower than the MongoShell), but something I've done that allowed me to run an aggregation in PHP (due to the timeout problems) is changing the way I invoked the aggregation.

Hope this helps someone that reaches this page because of the timeout problems, like I did!

Instead of $data->aggregate($ops) I ran the following equivalent to your case:

$db->command(
    array('aggregate' => 'data', 'pipeline' => $ops),
    array('timeout' => 100000000)
)

Notice that you must run the command over the $db and not your collection.