Why does Elasticsearch replicas get out of sync?

Question

Web development

I have two servers. The index size is not large (hundreds of thousands of documents), so I don't need shards.
But I need replicas for load balancing.
Not much data is added/updated per day - hundreds of documents.
But sometimes a problem arises, with one request to the first and second servers, different data is issued.
Config on the first node

cluster.name: mycluster
node.name: node-1

node.master: true
node.data: true

network.host: 0.0.0.0
transport.tcp.port: 9300
http.port: 9200

discovery.zen.ping.multicast.enabled: false
discovery.zen.ping.unicast.hosts: ["192.168.0.1", "192.168.0.1"]

On the second node, the config is the same except for two lines

node.name: node-2
node.master: false

curl -XGET ' localhost:9200/_cluster/health?pretty=true '

{
  "cluster_name" : "mycluster",
  "status" : "green",
  "timed_out" : false,
  "number_of_nodes" : 2,
  "number_of_data_nodes" : 2,
  "active_primary_shards" : 10,
  "active_shards" : 20,
  "relocating_shards" : 0,
  "initializing_shards" : 0,
  "unassigned_shards" : 0,
  "delayed_unassigned_shards" : 0,
  "number_of_pending_tasks" : 0,
  "number_of_in_flight_fetch" : 0,
  "task_max_waiting_in_queue_millis" : 0,
  "active_shards_percent_as_number" : 100.0
}

Write requests are sent to both the master and the replica.
1. Perhaps you should force the number_of_shards and number_of_replicas parameters, and re-create the index? Or it shouldn't matter.

index.number_of_shards: 1
index.number_of_replicas: 1

And I don't quite understand what numbers should be indicated here? 1 shard and 1 replica? What if there are 3 servers? - 1 shard and 2 replicas?
2. How can you understand that some kind of desynchronization has occurred?

Answer 1

I figured out what it was. It wasn't a matter of synchronization. And that the sort order was not defined. Therefore, when requested, different nodes produced different results. And even with a request to one node, the result was different (apparently, elasxerch itself does the balancing of requests). Added sorting by id, the problem was solved.