MongoDB cluster on Kubernetes - /opt/percona/ line 522: exec: numactl --interleave=all: not found

Hi, I have a kubernetes cluster which is running on Virtualbox and I am setting up a 3 node MongoDB replica set using Percona MongoDB Operator. Everything was working just fine with some very minor changes on the cr-minimal.yaml. I recently deleted the PerconaServerMongoDB and tried to recreate it and I am getting this error,

kubectl logs minimal-cluster-rs0-0
Defaulted container "mongod" out of: mongod, mongo-init (init)
+ exec 'numactl --interleave=all' mongod --bind_ip_all --auth --dbpath=/data/db --port=27017 --replSet=rs0 --storageEngine=wiredTiger --relaxPermChecks --clusterAuthMode=keyFile --keyFile=/etc/mongodb-secrets/mongodb-key --enableEncryption --encryptionKeyFile=/etc/mongodb-encryption/encryption-key --wiredTigerIndexPrefixCompression=true --tlsAllowInvalidCertificates
/opt/percona/ line 522: exec: numactl --interleave=all: not found

I am not sure where to look for the error.

My cluster config is

kubectl get nodes
NAME              STATUS   ROLES           AGE    VERSION
flatcar-worker1   Ready    control-plane   128d   v1.28.0
flatcar-worker2   Ready    <none>          128d   v1.28.0
flatcar-worker5   Ready    <none>          106d   v1.28.0

and the manifest I created it very simple

kind: PerconaServerMongoDB
  name: minimal-cluster
  crVersion: 1.16.0
  image: perconalab/percona-server-mongodb-operator:main-mongod6.0
  allowUnsafeConfigurations: true
    apply: disabled
    schedule: "0 2 * * *"
  # secrets:
  #   users: minimal-cluster
    users: my-cluster-name-secrets

  - name: rs0
    size: 3
        storageClassName: longhorn
            storage: 3Gi

    enabled: false

      size: 1
          storageClassName: longhorn
              storage: 3Gi

      size: 1

The replica set was just working fine and after deleting, during the recreation I am getting this error. Any pointers will be really helpful.

Thanks and Regards,

@athreyavc please do not use main branch. It is under development right now and from time to time we can see some problems with it. Please use PSMDBO v1.15.0.

@athreyavc main branch should work now.

1 Like

I did use 1.15.0 and it is working.

We experience the same issue with operator 1.15.0 and image version 6.0.13.
6.0.12 is working fine.

We are also experiencing this same issue with operator 1.15.0, image 6.0.9 and 6.0.12 both work fine but 6.0.13 and 6.0.14 have this issue. Is there any way we can re-open this ticket?

@Zarch @LTSmash Please see our official release notes Percona Operator for MongoDB 1.15.0 (2023-10-09) - Percona Operator for MongoDB

As you can see, psmdb 1.15.0 supports MongoDB vers 4.4.24, 5.0.20, and 6.0.9. If you want to use the latest PSMDB docker images (numactl was added into the latest ver of docker images and it causes this problem ) you need to wait for 1.16.0 operator because this version will have a fix in entrypoint.

It would be nice if the “Versions compatibility” listed patch versions then as it only lists major and minor, not patch. This ends up leading me to believe any 6.0.X should be supported by 1.15.0.