Skip to content

Improve slurm handling of maintenance reservations #3572

@ekouts

Description

@ekouts

Sometimes in CSCS we need to run --distribute together with a reservation that has enabled the flag MAINT. Even if we run with the RFM_IGNORE_REQNODENOTAVAIL set to True, we are getting failures for nodes that are not down and part of the selected maintenance reservation. Reframe always assumes that the node is down when its state is MAINTENANCE.

Metadata

Metadata

Assignees

Type

Projects

Status

Todo

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions