Percona MongoDB operator is causing high number of connections to the database

[Sai_krishna]

Hi all,

Percona operator is causing connection spike on the database. Here is the sample connection. Can we throttle the connections from operator?

type: 'op',
  host: 'mgdb-rs0-1:27017',
  desc: 'conn326588',
  connectionId: 326588,
  client: '10.0.0.111:54304',
  clientMetadata: {
    driver: { name: 'mongo-go-driver', version: 'v1.12.1' },
    os: { type: 'linux', architecture: 'amd64' },
    platform: 'go1.20.9'
  },
  active: true,
  currentOpTime: '2024-03-07T17:28:38.017+00:00',
  threaded: true,
  opid: 55231681,
  secs_running: Long("9"),
  microsecs_running: Long("9012110"),
  op: 'command',
  ns: 'admin.$cmd',
  command: {
    hello: 1,
    helloOk: true,
    topologyVersion: {
      processId: ObjectId("65e7659bc3f54ff574dfe7d3"),
      counter: Long("19")

[Sergey_Pronin]

Hey @Sai_krishna ,

thanks for bringing this up.
The issue right now is that we reconcile the psmdb object in Operator every 5 seconds, even there are no changes. This creates a connection and sends some pings to the cluster.

The best solution here would be to improve this reconcilation on the Operator level, not throttling it on DB side.

Meanwhile, could you please share if it impacts your database in any way?

[Pim]

I’m experiencing the same with Percona Operator for MongoDB 1.21.0 on AWS EKS 1.34.
After starting the Operator it keeps creating connections until it reaches OOM and dies. In my case this is around 8000 connections. It creates about one connection per second.

Please let me know what info I can provide to help find a solution.
I’ve created a seperate post here: https://forums.percona.com/t/percona-operator-for-mongodb-endlessly-spawning-connections-until-oomkilled