Skip to content

cluster health: take into account unschedulable nodes in other graphs #465

@tiborsimko

Description

@tiborsimko

Current behaviour

Sometimes one needs to cordon nodes in the cluster, and the health page may look like this:

Image

The problem is that we indicate a lot of available job slots (415) even though most of the nodes that are available to run jobs are cordoned.

Expected behaviour

If health page could read the status of nodes and based on their labels (node_label_runtimejobs: reana.io/system=runtimejobs) could discover how many jobs are available to run user jobs, we colud preset actual real number and colour some of those job slots as red.

Ditto for workflows (node_label_runtimebatch: reana.io/system=runtimebatch) and notebooks (node_label_runtimesessions: reana.io/system=runtimesessions) if we could analyse the node labels. We could then make better split between what nodes are available for each purpose.

Metadata

Metadata

Labels

No labels
No labels

Type

No type

Projects

Status

In work

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions