Skip to content

Glob based worker node job affinity #1508

@LNSD

Description

@LNSD

When deploying a dataset sync workload to the cluster via the Admin API, you can pass an optional "node id" parameter to pin a job to a worker. We can extend that behavior with a "node-id" glob.

For example, deploying to worker-raw-eth-* would match one of wokrer-raw-eth-mainnent and worker-raw-eth-l2s.

Note that:

  • xxxx-* is not a valid NodeId, an enum named something like GlobOrNodeId should be used to detect the param type.
  • The scheduler should randomly select one of the nodes with names matching the glob.

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions