Compute is in preview, and the
v1alpha API can change.Request access
Compute is enabled for each project separately, and Datum approves each request manually. Until a project has access, everydatumctl compute command except access and help stops and asks you to request access. This includes build.
To request access for the active project, follow these steps:
-
To check the current state, run the following command:
The
Statusline shows a state such asNot requested,Pending approval,Active,Denied, orRevoked. -
To submit a request and wait for a decision, run the following command:
Replace
MESSAGEwith a short description of what you plan to run. The message helps Datum review your request. -
Wait until the
Statusline showsActive. You can then deploy workloads.
datumctl compute access request --renew. For all access options, see Compute commands.
Quotas
Your project’s compute quota limits how many workloads and Instances you can run, and how much vCPU and memory they use. Datum sets the quota. You can’t change it yourself. Each Instance uses the following quota:- 1 Instance
- 1 vCPU
- 2 GiB (2,048 MiB) of memory
general-purpose Instance and a unikernel Instance use the same amount.
Check your quota
To see your project’s limits and usage, run the following command:--constrained.
If your quota is used up
If your project doesn’t have enough quota for a new Instance, Datum doesn’t start it. The Instance showsPending (quota exceeded) in datumctl compute instances and reports QuotaExceeded, and the workload reports QuotaNotGranted. To let the Instance start, do one of the following:
- Lower the number of Instances with
datumctl compute scale WORKLOAD_NAME --min=COUNT. - Delete workloads that you no longer need.
- Ask for more quota by contacting support@datum.net.
QuotaNoBudget, or another reason that starts with Quota, your project’s quota isn’t set up correctly. The workload exists, but its Instances don’t start. Contact support, and Datum fixes the problem.
Limits
The following table lists fixed limits:Limitations
The Compute preview has the following limitations. For limitations of each runtime class, see Run a container image and Build and deploy a unikernel. For limitations of the ALB, see Publish a workload.Runtime and sizing
- One instance type is available, and you can’t set CPU or memory for each container.
- Persistent disks, raw devices, GPUs, and custom virtual machine images aren’t available. Files written at runtime are lost when Datum replaces an Instance.
- Instances with more than one container aren’t supported. Run one container per workload.
- Health probes, such as readiness and liveness checks, aren’t available. An Instance is ready when its containers are running.
- You can’t suspend, resume, or snapshot Instances, or open a shell in them or run commands in them.
- A workload’s runtime class and network can’t change after you create it.
Placement and scaling
- Compute runs in
us-central-1(Dallas),us-east-1(Ashburn), andus-west-1(San Jose). - Workloads don’t scale automatically, and they can’t scale to zero. Don’t set
scaleSettings.maxReplicasorscaleSettings.metrics. - Rollouts replace Instances one at a time in each location, with no option to add extra Instances during the rollout.
Networking
- During the preview, Instances have private IPv6 addresses only. Private IPv4 addresses and public IP addresses are a preview limitation.
- Outbound internet access and IPv4 destinations are on the roadmap. Image pulls aren’t affected.
- Private DNS names for Instances are on the roadmap.
- Firewall rules to filter traffic between Instances on a network are on the roadmap.
- Public access is HTTP and HTTPS only, served through Datum’s ALB. Raw TCP and UDP exposure is a preview limitation.
Operations
datumctl computehas no command to read Instance logs.- Per-Instance CPU and memory metrics aren’t available.
What’s next
- To deploy your first workload, see Compute quickstart.
- To check status and fix problems, see Manage and troubleshoot workloads.