I'm trying to understand how the shards will increment indexStats based on a question in MongoDB University. I got the answer, but still struggling to understand it.
Here's the question:
You have a sharded cluster with 4 shards. The products collection is sharded on { name : 1, dateCreated : 1 }. There are 4 chunks, one on each server, and here is the range of the shard key:
s1: { name: MinKey, dateCreated: MinKey} - { name: "e", dateCreated: MinKey}
s2: { name: "e", dateCreated: MinKey} - { name: "g", dateCreated: MinKey}
s3: { name: "g", dateCreated: MinKey} - { name: "m", dateCreated: MinKey}
s4:{ name: "m", dateCreated: MinKey} - { name: MaxKey, dateCreated: MaxKey}
You have the following indexes:
{ _id: 1 }
{ name: 1, dateCreated: 1 }
{ price : 1 }
{ category: 1 }
You perform the following query:
db.products.find( { name : { $in : [ "iphone", "ipad", "apple watch" ] } } )
Which of the shards will increment their value of $indexStats for an index as a result of this query?
Answer is: s1 and s3
The first thing to note is that the query uses the shard key, and that it uses just the index prefix. Next, we note that:
"iPhone" is in shard s3
"iPad" is in shard s3
"Apple Watch" is in shard s1
Therefore, it'll be a targeted query, and will hit shards s1 and s3.
While it wasn't part of the question, it increments the {name : 1, dateCreated : 1} index on those shards.
So how do you find that iPhone and iPad go to s3 shard but not to the s2? If anyone could briefly explain this to me, I'd truly appreciate it.
Thanks!
Shard key ranges are cached in the mongoses from the config server , when you query the cluster from mongoses using the shard key the mongoses already know to which shard to forward the query so only in those shards the $indexStats value for {name:1,dateCreated:1} will increment.