Hybrid Deployment
Deploy Valohai workers and storage to your Oracle Cloud Infrastructure (OCI) tenancy while Valohai manages the application layer at app.valohai.com.
What Gets Deployed
The Compute and Data Layer of Valohai can be deployed to your OCI tenancy. This enables you to:
Use your own Compute instances to run machine learning jobs
Use your own OCI storage for training artifacts (trained models, preprocessed datasets, visualizations)
Access databases and data warehouses directly from workers inside your network
Valohai doesn't have direct access to the compute instances that execute machine learning jobs. Instead, it communicates with a Redis queue in your tenancy that's responsible for storing the job queue, job states, and short-term logs.
Prerequisites
From Valohai:
(Optional)
queue_address- DNS name for your queue (e.g.,something.vqueue.net)
From your OCI account:
Admin access to your OCI tenancy
A user for programmatic (API) access
Region selected (consider Compute and GPU availability)
Ability to run a small Compute instance for the Redis queue (set up in Step 1)
Step 1: Set Up the Redis Queue
Valohai uses a Redis queue in your tenancy for job distribution. At a high level, you'll start an instance, open the required ports in your VCN security list, fix Ubuntu's software firewall, and start the Redis service.
1. Create the instance
Create a Compute instance on OCI using a standard Ubuntu image (version 22.04 or newer recommended).
2. Open ports in the VCN
Make sure the VCN's security list accepts ingress connections on ports 63790 (Redis) and 80 (Let's Encrypt HTTP challenge).
3. Connect to the instance
Use SSH to connect to the instance.
4. Configure the software firewall
Open the Redis port 63790 and port 80 for the ACME certbot challenge by adding the following lines to /etc/iptables/rules.v4, after the existing SSH rule (-A INPUT -p tcp -m state --state NEW -m tcp --dport 22 -j ACCEPT):
Apply the changes:
5. Start Redis as a service
Run the helper script, replacing <redis_password> and <subdomain> with your own values:
Note: Keep the subdomain and password — the resulting queue URL is the Queue URL you'll provide in Step 3.
💡Using the <subdomain>.vqueue.net DNS name provided by Valohai is optional. You can also use your own one. In this case, make sure to let your Valohai contact know which one you used.
6. Check the service status
Test the Setup
💡 If you used the vqueue.net DNS name provided by Valohai, before testing make sure to inform your contact about the public IP of the queue machine.
From your local machine, connect to the queue and send a PING:
A PONG response confirms the queue is reachable.
Step 2: Prepare Your OCI Credentials
Collect the following details for the user Valohai will use to connect to and scale your OCI compute.
For detailed instructions on how to find these values, refer to OCI's documentation on required keys and OCIDs.
User OCID – The OCID of the user that will make API calls (starts with
ocid1.user.oc1..).Fingerprint – The fingerprint of the public key associated with the user.
Private Key – The private key file or its contents.
Passphrase – The passphrase for the private key, if it is encrypted.
Tenancy OCID – The OCID of your tenancy (starts with
ocid1.tenancy.oc1..).Region – Your OCI region, for example
eu-stockholm-1. See OCI Regions for reference.
Important: The private key and its passphrase are sensitive credentials. Share them with your Valohai contact using your organization's secure communication method (for example, a password-protected document or encrypted email) — never in plain text.
Step 3: Define Compute Configuration
Specify the compute environment you want to use with Valohai. Provide the following details for each environment.
Required Settings
Compartment OCID – The compartment under which instances will be created.
Shapes – The instance types you want to use. For flexible shapes, also include the desired CPU and memory configuration.
Disk Size – Size of the root volume in GB.
Queue URL – A Redis queue URL used for job distribution, format:
Subnet OCID – The subnet the instances should connect to.
Optional Settings
Freeform Tags – Any additional tags you'd like to attach to the instances.
Image – Either the OCID of a custom image or the desired OS and version. By default, Valohai uses the latest supported Ubuntu image.
SSH Public Key – A public key for optional SSH access.
Disable Public IP – Indicate if a public IP should not be assigned. By default, one is assigned.
Summary
Collect the following information to send to Valohai.
Credentials:
User OCID:
____________Fingerprint:
____________Private Key:
____________Passphrase (if encrypted):
____________Tenancy OCID:
____________Region:
____________
Compute Configuration:
Compartment OCID:
____________Shapes (and CPU/memory for flexible shapes):
____________Disk Size (GB):
____________Queue URL:
____________Subnet OCID:
____________
Optional:
Freeform Tags:
____________Image (OCID or OS/version):
____________SSH Public Key:
____________Disable Public IP (yes/no):
____________
Share this information with your Valohai contact using your organization's secure communication method.
Next Steps
After Valohai confirms your environment is configured:
1. Verify the setup
Log in to app.valohai.com
Check that OCI environments appear in your organization
Run a simple execution to verify workers launch correctly
2. Configure additional resources
Add existing OCI storage as data stores
Set up private Docker registries
Configure access to your databases
Getting Help
Valohai Support: support@valohai.com
Include in support requests:
Tenancy OCID
Region
Error messages or logs
Steps already attempted
Last updated
Was this helpful?