An easy-to-use, flexible, and performant workload orchestrator that can deploy a mix of microservice, batch, containerized, and non-containerized applications.
Nomad is a simple and flexible workload orchestrator to deploy and manage containers (docker, podman), non-containerized applications (executable, Java), and virtual machines (qemu) across on-prem and clouds at scale.
Nomad provides several key features:
- Deploy Containers and Legacy Applications: Nomad’s flexibility as an orchestrator enables an organization to run containers, legacy, and batch applications together on the same infrastructure. Nomad brings core orchestration benefits to legacy applications without needing to containerize via pluggable task drivers.
- Simple & Reliable: Nomad runs as a single binary and is entirely self contained - combining resource management and scheduling into a single system. Nomad does not require any external services for storage or coordination. Nomad automatically handles application, node, and driver failures. Nomad is distributed and resilient, using leader election and state replication to provide high availability in the event of failures.
- Device Plugins & GPU Support: Nomad offers built-in support for GPU workloads such as machine learning (ML) and artificial intelligence (AI). Nomad uses device plugins to automatically detect and utilize resources from hardware devices such as GPU, FPGAs, and TPUs.
- Federation for Multi-Region, Multi-Cloud: Nomad was designed to support infrastructure at a global scale. Nomad supports federation out-of-the-box and can deploy applications across multiple regions and clouds.
- Proven Scalability: Nomad is optimistically concurrent, which increases throughput and reduces latency for workloads. Nomad has been proven to scale to clusters of 10K+ nodes in real-world production environments.
- HashiCorp Ecosystem: Nomad integrates seamlessly with Terraform, Consul, Vault for provisioning, service discovery, and secrets management.
dynamic host volumes: Fixed a bug where users with host-volume-delete in one namespace could delete claims from another namespace [GH-28205]
IMPROVEMENTS:
cli: Add a -kv-path flag to nomad setup vault to configure the Vault KV mount used by the generated workload policy [GH-28183]
cli: Added -json and -t options to the operator autopilot get-config command. [GH-27991]
client: Add tunable for Vault default lease duration on templates for paths without leases. [GH-28199]
consul: Allow service, template, and connect blocks to fallback to the Nomad client agent's Consul token if workload identity is unavailable [GH-28106]
driver: Added optional Init function for task driver plugins [GH-28104]
driver: Added optional Shutdown function for task driver plugins [GH-28102]
scheduler: Stop failed allocations first when downscaling a task group [GH-27971]
DEPRECATIONS:
agent: Unauthenticated server join via the CLI or API is deprecated. [GH-28176]
BUG FIXES:
api: allow using WI tokens on plan endpoint [GH-28139]
cli: Fixed a bug where complex HCL variables passed via -var flag could not be edited in the web UI [GH-28138]
client: Fixed a bug where a client could panic after an alloc is GC'd [GH-28185]
docker: Enforce allowed_modes or allow_privileged requirement to set host namespace modes in task [GH-28190]
dynamic host volumes: Fixed a bug where allocations claiming host volumes with the per_alloc flag would not prevent the volume from being deleted [GH-28198]
metrics: expired metrics are now periodically cleared from the Prometheus sink even if no collection occurs [GH-28170]
scheduler: Fixed a bug where a node could be marked feasible for a task group requesting multiple host volumes when a satisfied sticky volume request short-circuited the checks for the remaining requests [GH-28097]
scheduler: Fixed a bug where setting sticky on a static host volume could fail the evaluation instead of being rejected during feasibility checking [GH-28097]
scheduler: keep draining batch alloc counted when node is re-enabled [GH-28018]
task runner: Improve the memory management for secrets [GH-28140]
ui: Fixed a bug where jobs that share a ModifyIndex (for example, several jobs rescheduled in a single Raft transaction after a node failure) were omitted from the jobs page and the /v1/jobs/statuses endpoint [GH-28132]
ui: fixes an issue where streaming task logs would error [GH-28137]
