Get AuthorizationException when trying to update variables of a started workflow

[jruizaranguren]

I'm trying to set some workflow variables just after starting a workflow with this code:

clients = OrkesClients(configuration=config)
executor = WorkflowExecutor(config)
runner = clients.get_workflow_client()

request = StartWorkflowRequest(
            name="flow", correlation_id="corr_id", version=1)
flow_id = runner.start_workflow(request)
runner.update_variables(flow_id, variables={"myglobal": "global"})

But I'm getting a 404 error, which is returned as result of an internal ApiException raised from conductor/client/http/rest.py. It seems it is unable to reach /workflows/{workflow_id}/variables:

Could this be something related to the workflow still starting execution?

[jruizaranguren]

The same goes when trying to get a workflow status via WorkflowExecutor or OrkesWorkflowClient. It results in a /api/workflow/1f3e63cfef74495e888993df6ac01b71/status not found exception.

[IgorChvyrov-sm]

Hi @jruizaranguren

Thanks for reporting this issue! To help us investigate, could you please share:

1. The version of the conductor-python library you're using
2. The version of Conductor you're running

[nthmost-orkes]

Reproduced this. update_variables() reliably 404s against the OSS Conductor server — and it's not a config or auth problem.

The SDK calls POST /workflow/{workflowId}/variables, but that endpoint doesn't exist in OSS Conductor. It's Orkes-only (WorkflowResource.updateWorkflowState). So any OSS user calling update_variables() gets a bare 404 with no hint as to why.

For what it's worth, get_workflow_status() and get_workflow() both work fine against OSS — so the status 404 mentioned above was likely a separate transient issue, not the same root cause.

Keeping this open. The fix on our side is to have update_variables() raise a clear "not supported on OSS Conductor server" error instead of a confusing 404, so it's obvious this is an Orkes feature rather than a bug.