Query is very slow while adding le function for float predicate in filter

Issues Dgraph
4

@zedzhu I haven’t had time to set up an HA cluster to support this query yet, but in the meantime could you try the following query? One thing about DQL is that it really just as much a set query language as it is a graph query language. The idea of the query below is to reduce the initial set against the most possible restricted predicate. Then keep reducing the set by additional predicates. You can use the num_uids in the metrics in the extended results as a guide.

{
  CID_SET as var(func: eq(cid3_1, "717"))
  
  BRAND_SET as var(func: uid(CID_SET)) @filter(eq(brand_1, "GODEX"))

  PRICE_SET as var(func: uid(BRAND_SET)) @filter(le(price_1, 2000))
    
  q(func: uid(PRICE_SET), first: 1000, offset: 0) {
		sku_id_1
		cid3_1
		price_1
  }
}
1 Like