2023-01-10 17:50:31 +01:00
|
|
|
---
|
|
|
|
|
title: "NFD-Topology-Garbage-Collector"
|
|
|
|
|
layout: default
|
|
|
|
|
sort: 6
|
|
|
|
|
---
|
|
|
|
|
|
|
|
|
|
# NFD-Topology-Garbage-Collector
|
|
|
|
|
{: .no_toc}
|
|
|
|
|
|
|
|
|
|
---
|
|
|
|
|
|
|
|
|
|
NFD-Topology-Garbage-Collector is preferably run as a Kubernetes deployment
|
|
|
|
|
with one replica. It makes sure that all
|
2023-01-11 17:23:36 -08:00
|
|
|
[NodeResourceTopology](custom-resources.md#noderesourcetopology)
|
2023-01-10 17:50:31 +01:00
|
|
|
have corresponding worker nodes and removes stale objects for worker nodes
|
|
|
|
|
which are no longer part of Kubernetes cluster.
|
|
|
|
|
|
|
|
|
|
This service watches for Node deletion events and removes NodeResourceTopology
|
|
|
|
|
objects upon them. It is also running periodically to make sure no event was
|
|
|
|
|
missed or NodeResourceTopology object was created without corresponding worker
|
|
|
|
|
node. The default garbage collector interval is set to 1h which is the value
|
|
|
|
|
when no -gc-interval is specified.
|
|
|
|
|
|
|
|
|
|
## Topology-Garbage-Collector Configuration
|
|
|
|
|
|
|
|
|
|
In Helm deployments,
|
|
|
|
|
(see [Topology Garbage Collector](../deployment/helm.md#topology-garbage-collector-parameters)
|
|
|
|
|
for parameters). NFD-Topology-Garbage-Collector will only be deployed when
|
|
|
|
|
topologyUpdater.enable is set to true.
|
