Skip to content

agent: fatal error: concurrent map writes #12460

@mvisonneau

Description

@mvisonneau

Bug report

Scaling out the amount of nodes of a kubernetes cluster, some of the cilium agent (~18%) failed on the first run with the following error : fatal error: concurrent map writes. They all seemed to be working fine after the very next automated restart of the agent process.

at first sight, the error seems quite similar to #8278

General Information

I am happy to provide the sysdump if necessary as it won't fit on this GitHub issue (>10MB).

How to reproduce the issue

I am very uncertain about which environment specifics may be in cause, all I can tell for now is that it never happened until today as I attempted to scale from 5 nodes to 75. All t3a.xlarge in a single AZ. Scaled out via the cluster-autoscaler after increasing the amount of replicas of an httpbin deployment.

Metadata

Metadata

Assignees

Labels

kind/bugThis is a bug in the Cilium logic.kind/community-reportThis was reported by a user in the Cilium community, eg via Slack.

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions