Continuous admin.$cmd queries in slowest operations

Hello,
we are facing one strange issue in the Realtime Slowest operations. It seems that for every docker that connects to mongoDB we have one or more continuous admin.$cmd commands that are reading some thousands of documents.
All our dockers are NodeJS projects accessing the database via mongoose. One sample of the JSON RESPONSE is:
op: command
locks: Object
lockStats: Object
waitingForFlowControl: false
host: xxxx-shard-00-02-vckat.azure.mongodb.net:27017
microsecs_running: 8432221
currentOpTime: 2021-06-09T12:02:02.646+00:00
secs_running: 8
waitingForLock: false
flowControlStats: Object
type: op
active: true
opid: 34729718
ns: admin.$cmd
numYields: 0
waitingForLatch: Object
timestamp: Object
$date: 2021-06-09T12:01:54.314Z
captureName: FutureResolution
connectionId: 55945
clientMetadata: Object
driver: Object
name: nodejs|Mongoose
version: 3.6.6
os: Object
type: Linux
name: linux
architecture: x64
version: 5.4.0-1032-azure
platform: 'Node.js v10.15.3, LE (unified)
version: 3.6.6|5.12.5
desc: conn55945
client: xx.xx.xx.xx:49380
query: Object
$db: admin
ismaster: true
maxAwaitTimeMS: 10000
topologyVersion: Object
processId: Object
$oid: 60bd2bb49561137ee56a1868
counter: 8
$clusterTime: Object
clusterTime: Object
$timestamp: Object
t: 1623235570
i: 2
signature: Object
hash: Object
$binary: Object
base64: /YEfTf0O9ucqS31QdgoFKDLo0qg=
subType: 00
keyId: 6914106613016035000

We are using MongoDB version 4.4.6.

Do you have an idea on what type of activity the system is performing?

Thanks