Oracle Cloud Infrastructure Documentation

Known Issues for Roving Edge

The following known issue is identified in Roving Edge.

Public SDK and public Terraform not available for data sync at this time

At this time, public SDK and public Terraform versions are not available for data sync.

Details
While public SDK and Terraform are not currently supported, there are supported versions of available.
Workaround
Use the following supported SDK and Terraform versions for data sync operations:

edgepeerconnection can remain ACTIVE when peerserialnumber is incorrect

If you enter an incorrect peerSerialNumber while creating an EdgePeerConnection, the create job can remain in the ACTIVE state indefinitely and requires manual cancellation to recover.

Details
When EdgePeerConnection is created with incorrect peer connection parameters—most commonly an invalid peerSerialNumber—the workflow can stall while waiting for the tunnel to come up, leaving the associated work item in an Active state indefinitely with progress such as “checking if tunnel is connected.” Depending on where the failure surfaces, one rack may fail later during trust exchange with an explicit serial-number-related error, while the other may never time out and remains stuck in Creating/Active.
Workaround
Manually cancel the Active work item. After cancellation, delete the incomplete EdgePeerConnection and recreate it using the correct peer parameters.

  1. Gather the ID of the stalled job using the getallactiveJobs command.

  2. Gather the Active work item within that job, using the show job id=<jobId> command.

  3. View the details of the work item using the showworkItem id=<workItemId> command.

  4. Cancel the active work item using the cancelworkItem id=<workItemId> command.

  5. Delete the EdgePeerConnection using the delete EdgePeerConnection id=<Id>

  6. Recreate the EdgePeerConnection using the correct peerSerialNumber.

Load Balancer session persistence not available for this release

At this time, Load Balancer session persistence is not supported for this release.

Details
Session persistence options, such as cookie-based or source IP-based persistence, are currently unavailable in the Load Balancer for this release.
Workaround
We're working on a fix for this issue.

VCN-related network resources become stuck

VCN network resources might get stuck in PROVISIONING or FAIL states, because of OVN access object pool exhaustion.

Details
You may encounter a “Northbound database access connection object” error when the OVN (Open Virtual Network) access object pool is exhausted. When this happens, API calls can repeatedly fail and VCN network resources may get stuck in PROVISIONING or FAIL states.
Workaround
Increase the pool size OVN Northbound database access connections from 32 to 64.

OKE cluster add-ons not supported at this time

At this time, enabling OKE add-ons might fail across multiple cluster configurations.

Details
Customers might see add-on enablement operations fail regardless of cluster type or node shape. This behavior has been observed on Flannel clusters with non-GPU worker node shapes and may impact other configurations as well.
When viewing the node POD status in the OKE namespace, you might see the OKE addon-controller restarting repeatedly. This is a known issue.
Workaround
We're working on a solution.

Yum package installs might fail when using OL7.9 image

At this time, yum operations might fail on PCA environments when using the uln-pca-Oracle-Linux-7.9-2025.07.21_0.oci image because of Oracle Linux 7 Extended Lifecycle Support (ELS) repositories being enabled by default.

Details
In the latest OL7.9 image, ELS repositories (for example, ol7_UEKR6_ELS) are enabled by default. These repos might resolve to yum.oracle.com, where the ELS content is not published, resulting in 404 Not Found errors for repository metadata (for example, repodata/repomd.xml) and causing yum to fail.
Workaround
Disable ELS repositories when running yum.

During a reboot or power cycle, the device status might report "Unexpected Error!"

Details

As of version 2.18, the following error might be displayed when you reboot or power cycle a device:

Version          : *** Unknown ***System Time UTC     : 2025-08-28 18:14:31
Boot Time                  : 2025-08-28 17:39:18
Device Serial Id           : Unexpected Error!
Unlock Status              : Unexpected Error!
Auto Unlock Status         : Unexpected Error!
Diag Mode                  : *** Unknown ***
Red API Services Status    : *** Unknown ***
Red HW Services Status     : *** Unknown ***
Press ENTER to return...
Workaround

Reboot the device until the error is no longer displayed. See Rebooting a Roving Edge Device.

An Oracle Linux 9 instance takes a while to boot

An Oracle Linux 9 instance might take longer than 4 minutes to boot.

Workaround

  1. In the instance, run the following command to check if the bootloader entry contains a netroot setting that's set to an iscsi target:

    grep "netroot=iscsi" /boot/loader/entries/*$(uname -r).conf

  2. If the command returns any result, remove the netroot option from the bootloader entries using this command:

    sed -i 's/netroot=iscsi:[^ ]\+ / /' /boot/loader/entries/*.conf