HashiCorp Nomad Memory Calculator
Nomad places tasks according to their memory reservation. With memory oversubscription enabled,
memory_max lets a task grow past that reservation, up to a hard limit, whenever its client has
memory to spare — the same idea as Kubernetes requests and limits. This calculator sizes both from real usage.
01 Usage data
Which query should I paste?
With telemetry's publish_allocation_metrics and prometheus_metrics on, use a range query over a representative period (a week is a good default) for one task:
nomad_client_allocs_memory_usage{job="web", task="api"}
Paste the Prometheus UI table output, the HTTP API JSON, or a Grafana CSV export.
Enter observed usage directly, in MiB (Nomad's "MB").
02 Workload profile
Advanced: margins and defaults
Values the result relies on that are our own defaults, or vendor defaults your setup may change. Each “?” says how one moves the result; the guide says where it comes from.
03 Sizing gauge
04 Nomad job (task resources)
—05 How this was derived
How every number is calculated, with sources and assumptions (opens in a new tab) ↗
Result looks wrong? Report it
memory, memory_max and oversubscription in short
memory is the memory the task requires, in MB. Without memory_max it is also the hard limit; with it set, memory becomes the reservation and memory_max the hard limit. Nomad's own advice is the typical usage for memory and memory_max to absorb unexpected load spikes. Nomad docs
So the calculator sizes memory from the average and memory_max from the peak, with margins that are MemoryLimit's own defaults, not vendor figures; you can change them under Advanced.
memory_max only works when memory oversubscription is enabled in the scheduler configuration. It is off by default and enabled once per cluster. Without it, memory is the hard limit, so a task sized from the average would be killed at that value; the note under the result says to set memory to the memory_max value in that case. Nomad docs
Common input mistakes
- Filtering by job only: filter by both job and task, or another job's task of the same name is mixed in.
- Pasting from a cluster whose telemetry has
publish_allocation_metricsoff: Nomad publishes allocation metrics only when it is on. - A window that misses the busiest times:
memory_maxcomes out too low.
Full derivation, warnings and every assumption: how the Nomad calculator works. The shared formula and margin tables: the sizing model.
Related calculators: