Skip to content

Memory leak in calico-felix (KDD Syncer) #1057

@tmjd

Description

@tmjd

It seems there is a memory leak in calico-felix. After leaving a cluster running for a day I checked on memory usage with top and noticed that calico-felix was reported as using 30% on my 2 nodes and 15% on the master. I restarted one of the calico-node pods (kubectl delete -n kube-system pod <pod-name>) and then checked memory usage and calico-felix reported 0.8% (2 hours later it is at 4.4%).

Expected Behavior

I would not expect 30% memory usage on a host with 4GB Ram.

Current Behavior

After a day of almost no usage calico-felix took 30% of the memory.

Possible Solution

Eh

Steps to Reproduce (for bugs)

  1. Bring up K8s cluster with Kubeadm
  2. Deploy Calico with http://docs.projectcalico.org/v2.5/getting-started/kubernetes/installation/hosted/rbac-kdd.yaml and http://docs.projectcalico.org/v2.5/getting-started/kubernetes/installation/hosted/kubernetes-datastore/calico-networking/1.7/calico.yaml
  3. Wait a day
  4. Check top on the hosts.

Context

Your Environment

  • Calico version: v2.5.0
  • Orchestrator version (e.g. kubernetes, mesos, rkt): K8s 1.7.4
  • Operating System and version: Ubuntu 16.10

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions