Hi,
I have a small (~38 megabytes) search index with a relatively simple query, however it is not performant with some queries with customer accounts that have a large amount of data indexed. The search will take longer than 60 seconds, which times out the HTTP request serving the frontend.
I’m trying to remove the $sort operation from the aggregation as recommended here: https://www.mongodb.com/docs/atlas/atlas-search/performance/query-performance/#-sort-aggregation-stage-usage
However, I’m a bit confused on how to use $near within the $search operation.
For context, I’m trying to sort based on a date field (createdAt), and the results should always be in descending order.
But I’m not sure if $near should be within a should clause or filter clause, and I’m also unsure how to use the pivot option to sort in a descending order:
const shopName = "test-inc";
const query = 'Smith';
const size = 25;
const page = 0;
const aggregation = [
{
$search: {
index: "checks",
compound: {
$filter: {
{
phrase: {
query: shopName,
path: "shopName",
},
},
},
$should: [
{
text: {
query: searchTerm,
path: [
"firstName",
"lastName",
"phone",
"email",
"orderId",
],
},
},
{
near: {
path: "createdAt",
origin: new Date(),
pivot: 86400000 * 7,
score: { boost: { value: 999 } },
},
},
],
}
},
{
$skip: parseInt(page) * parseInt(size),
},
{
$limit: parseInt(size),
},
{
$facet: {
customers: [{
$project: {
createdAt: 1,
name: 1,
email: 1
}
}],
},
{
$set: { meta: "$$SEARCH_META" },
},
];