Pools
Every machine runs are placed on, by pool: ec2, the instances orca launches for runs; kubernetes, a run's pods on the Kubernetes runner; and worker:<pool>, the machines of your own running orca worker. The one machine that runs the scheduler, this UI and Postgres is Control; Watchdog lists the rules that pause pipelines.
This scheduler's runner is local: it runs every run on its own machine and launches no instances, so there are none to list and no per-instance cost. The machine is Control; its price is the server's fixed price.
Pool: all worker:bgent 1/1
MachinesA row per machine of each pool runs are placed on: the EC2 instances orca launches, a Kubernetes run's pods, and the machines running orca worker for a worker:<pool> pool, the worker in the row's last column. Refreshed every 10 s. Docs ↗
| MachineThe machine: an EC2 instance's id, a Kubernetes run's pods, or the host name a self-hosted worker gave (its token's label until it has claimed). Click for its page. | PoolWhere a placement sends runs: ec2 (instances orca launches), kubernetes (a run's pods) or worker:<pool> (the machines running orca worker for that pool). Click to show only it. | Kindec2: an instance orca launched and pays for. k8s: a run's pods on the cluster. self-hosted: a machine of your own running orca worker. | Statelive: up and taking work. draining: the scheduler is letting it finish before it goes. gone: an instance that ended (listed for an hour), or a worker revoked, expired or not heard from for ORCA_STALE_SECONDS. Beside it, what it is doing. | ImageThe image tag its runs need: the pipeline's image on EC2 and Kubernetes, the worker's ORCA_IMAGE as it said. | CPUsIts vCPUs: the instance type's (or its sampler's count), the CPU a run's pods asked for, or the cores the worker said it has. | MemoryIts memory: the instance type's, or what the worker said it has. | RunsThe runs on it now. Click one for its page. | Last heartbeatWhen it was last heard from: an instance's last sample, a worker's last claim, a run's pods' last heartbeat. | CPUIts CPU in use at the last sample, and over the last ten minutes. Only EC2 instances are sampled for Ops. | WorkerWhat orca runs there: a run's head or a dista worker on EC2; orca worker, signed in with the token named (hover for when it expires); or a run's pods. |
|---|---|---|---|---|---|---|---|---|---|---|
bd410e935f57 | worker:bgent | self-hosted | live running | - | 8 | 11.7 GiB | 1439 1808 2685 | 0 s ago | - | orca worker token 1 · bgent |
Range: 24h 7d (UTC, by the hour)
Hosts over time
Hosts that ran during each hour, by role.
heads workers control
Runs per hour
Runs started and runs that ended each hour; failures (failed runs, and every failed step attempt, even one a retry fixed) are overlaid on the finished bar.
started finished failures (runs and attempts)
Fleet CPU
The vCPUs of all sampled hosts: busy, and available.
vCPUs busy vCPUs available vCPUs
Fleet network
Throughput summed over all sampled hosts.
in (MB/s) out (MB/s) MB/s