wip: Oracle Cloud support

[navaneeth-dev]

Adds Oracle Cloud Testing Support

Adds Oracle Cloud Testing Support.

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[navaneeth-dev]

private ipv4 & public ip is not exposed via afterburn in Oracle which is required for some tests. So I need to add it upstream. For now I will use systemd hack. Correct me if I am mistaken. @tormath1

https://github.com/coreos/afterburn/blob/main/docs/usage/attributes.md#metadata-attributes

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[navaneeth-dev]

I think my PR is ready. Let me know what else to implement.

=== RUN   cl.basic
=== RUN   cl.basic/ReadOnly
=== RUN   cl.basic/MachineID
=== RUN   cl.basic/CloudConfig
=== RUN   cl.basic/Script
=== RUN   cl.basic/SymlinkFlatcar
=== RUN   cl.basic/UpdateEngineKeys
=== RUN   cl.basic/ServicesActive
=== RUN   cl.basic/RandomUUID
=== RUN   cl.basic/Useradd
=== RUN   cl.basic/Microcode
=== RUN   cl.basic/PortSSH
=== RUN   cl.basic/DbusPerms
=== RUN   cl.basic/Symlink
2026-06-22T16:54:31Z platform/machine/oracle: leaving OCI instance ocid1.instance.oc1.ap-hyderabad-1.anuhsljrnkxuiqicotltjhtzulsbtt4d4uwwpkbd44tud7qbsd2mwmwdexlq running for SSH inspection
--- PASS: cl.basic (52.11s)
    --- PASS: cl.basic/ReadOnly (0.15s)
    --- PASS: cl.basic/MachineID (0.15s)
    --- PASS: cl.basic/CloudConfig (0.70s)
    --- PASS: cl.basic/Script (0.28s)
    --- PASS: cl.basic/SymlinkFlatcar (0.14s)
    --- PASS: cl.basic/UpdateEngineKeys (0.14s)
    --- PASS: cl.basic/ServicesActive (0.15s)
    --- PASS: cl.basic/RandomUUID (0.18s)
    --- PASS: cl.basic/Useradd (0.33s)
    --- PASS: cl.basic/Microcode (0.18s)
    --- PASS: cl.basic/PortSSH (0.14s)
    --- PASS: cl.basic/DbusPerms (0.54s)
    --- PASS: cl.basic/Symlink (0.14s)
PASS, output in _kola_temp/oracle-2026-06-22-2223-103411

[tormath1]

Thanks a lot, overally it looks good. I just have one request, usually the ore command is used to create and upload an image on the cloud provider. Is that possible to implement this?

It's then used like that: https://github.com/flatcar/scripts/blob/d2c217cb741debc9becda0bda86347319f17a65c/ci-automation/vendor-testing/stackit.sh#L24-L48

We first upload the image then we use the resulting image ID in the kola run command.

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[navaneeth-dev]

Done, tested ore image uploads.

➜ IMAGE_ID=$(./bin/ore oracle \
  --oracle-config-file="${ORACLE_CONFIG_FILE:-$HOME/.oci/config}" \
  --oracle-profile="${ORACLE_PROFILE:-DEFAULT}" \
  --oracle-compartment-id="${ORACLE_COMPARTMENT_ID}" \
  --oracle-bucket="${ORACLE_BUCKET}" \
  create-image \
  --board="${CIA_ARCH:-amd64}-usr" \
  --name="flatcar-oracle-upload-test-$(date +%s)" \
  --file="${ORACLE_IMAGE_NAME}"
)

echo "${IMAGE_ID}"
ocid1.image.oc1.ap-hyderabad-1.aaaaaaaairf3z42rr5qlebezgmxpyaitpyok7bm24inistcnujec7jiohbqa

Added retry also, using DefaultRetryPolicy of OCI SDK:

Retries 409 IncorrectState
Retries 409 LockConflict
Retries 429 TooManyRequests
Retries 5xx, except 501 MethodNotImplemented
Uses exponential backoff with jitter
Maximum wait per retry is about 30s + 1s jitter
Up to 8 attempts
Cumulative backoff is about 1.5 minutes
With eventual consistency handling, up to 9 attempts
Eventual consistency cumulative backoff is about 4 minutes

[tormath1]

Looks good, thanks! I'll give a try in the CI.

[tormath1]

Do we still need all of this now we use the OCI Retry mechanism?
If yes, let's just use the Mantle implementation:

$ go doc util.Retry
package util // import "github.com/flatcar/mantle/util"

func Retry(attempts int, delay time.Duration, f func() error) error
    Retry calls function f until it has been called attemps times, or succeeds.
    Retry delays for delay between calls of f. If f does not succeed after
    attempts calls, the error from the last call is returned.

[navaneeth-dev]

Checked the OCI Retry mechanism but it is too verbose, so using mantle's util.Retry for simpler approach. WaitForInstanceState wrappers are needed for nice error formatting.

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[tormath1]

I did a full run, here's the failing tests:

11:48:45  cl.ignition.v2.noop
11:48:45  cl.ignition.v1.noop
11:48:45  cl.ignition.misc.empty
11:48:45  cl.cloudinit.basic

cl.cloudinit.basic can be excluded as there is no cloudinit implementation for Oraclecloud in Flatcar.

For the three others, it should work - as Afterburn implementation is done. I think we're missing the SSH public key upload: Mantle generates public key, so we need to upload it and associate it the spawned instances.

As discussed on private, the ARM64 implementation needs some rework but we can merge this without for now.

Otherwise the rest looks good. Thanks a lot for this huge contribution.

[tormath1]

Suggested change

	sv(&kola.OracleCloudOptions.Region, "oraclecloud-region", "", "Oracle Cloud fingerprint")
	sv(&kola.OracleCloudOptions.Region, "oraclecloud-region", "", "Oracle Cloud region")

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.

[Copilot]

Copilot wasn't able to review this pull request because it exceeds the maximum number of files (300). Try reducing the number of changed files and requesting a review from Copilot again.