Telegraf adds irrelevant hostname from ShardHostData while gathering stats

[I-Hate-NONE]

Relevant telegraf.conf

[[outputs.influxdb]]
  urls = ["http://localhost:8086"]
  retention_policy= "autogen"
  write_consistency = "any"
  timeout = "30s"

[[inputs.mongodb]]
  servers = [ "mongodb://test:test@<replicasetnode>:27017/?connect=direct" ]
  gather_perdb_stats = true
  gather_col_stats = true
  gather_top_stat = true
  [inputs.mongodb.tags]
    "component_name" = "mgs"
    "ReplicaSet_name" = "test"

Logs from Telegraf

n/a

System info

master (https://github.com/influxdata/telegraf/tree/master)

Docker

No response

Steps to reproduce

I have a replica set with 3 nodes. I'm looking at the update_count metric to check the rate for a particular collection for a particular host. I have host label = my target host and hostname = the 2nd or the 3rd node in the replica set. Checked and debugged the code and saw that you:

1. Set the 'original' host's tag first:

   telegraf/plugins/inputs/mongodb/mongodb_server.go

   Line 38 in d118a07

   tags["hostname"] = s.hostname

2. Override it with another value:

   telegraf/plugins/inputs/mongodb/mongodb_data.go

   Line 459 in d118a07

   d.Tags["hostname"] = host.Name

3. In the end, when constructing a query for host = host1 I get two chart with (host=host1, hostname=host2) and (host=host1m hostname=host3) and I don't get a chart to see how many updates have been executed for my host1.

Expected behavior

Create a metric to show stats for a necessary host

Actual behavior

Two metrics are created; each looks like a separate counter; it's impossible to get the query to show the data for the dedicated primary.

Additional info

I may be missing something so would be grateful if you show me the proper way to set this up.