What's happening on Mjolnir right now?
Cluster activity is published in two places, and they answer slightly different questions.
Mjolnir Analytics — utilization and capacity#
Mjolnir Analytics publishes node- and capacity-level views, updated roughly hourly:
- Overview — current utilization at a glance
- Infrastructure
- Capacity — including utilization over time
This gives a structured live picture: how many nodes are up, aggregate CPU allocation, and queue depth by partition, without exposing anything about individual users, jobs, or projects.
Job usage statistics#
The Mjolnir website's Job usage statistics page shows job throughput over time:
Note Both views are aggregate. Neither shows other users' jobs, individual project usage, or account-level detail — that isn't published, and this page can't provide it either. If you need information about your own job specifically, see Monitoring jobs.
What this can tell you#
- Roughly how busy the cluster is right now, and how that's trended over time
- How many jobs are running versus queued, cluster-wide
- Whether nodes are available, down, or in maintenance
What this can't tell you#
Utilization is not a predictor of when your specific job will start. A cluster that looks busy doesn't mean your job can't start soon — a small request can still slot into a gap. A cluster that looks idle doesn't mean your job will start immediately, either — fairshare, QoS limits, and dependencies all factor into scheduling regardless of how free the cluster looks in aggregate.
Important Don't use utilization as a substitute for checking your own job. "The cluster looks quiet, my job should start any second" and "the cluster looks busy, something must be wrong with mine" are both unreliable. For your actual job, check its own pending reason — see Why is my job pending?.