Skip to content

Hydra Triage graph's stop message is stale: the OpenConnector command node IS installed #83

Description

@rubenvdlinde

The seeded Hydra Triage graph (lib/Repair/SeedHydraTriageFlow.php) has a command stop node whose message reads:

A label was proposed and recorded. No forge write was attempted: the OpenConnector-backed command node is not installed on this instance.

That is no longer true. GET /apps/openregister/api/flow/node-catalog on this instance returns 13 registered node types including openconnector.source-call and openconnector.synchronization-run (openconnector 0.3.3, files present in the container, registration gated only by a class_exists() guard on the event class which passes).

Why it matters: the message is the first thing anyone debugging the graph will read, and it points at the wrong cause. The actual blocker is credential scope — openconnector.source-call needs a Source, and a brokered GitHub Source needs a credential whose allowed_apps includes openconnector. Both brokered github credentials are scoped ["hermiq"], so the flow engine cannot make a forge call regardless of which nodes are registered.

I hit that credential blocker independently from the other direction before finding this graph; the two corroborate.

Suggested fix: replace the stop node with the real openconnector.source-call command step, or — if that is premature — reword the message to name the credential scope rather than a missing node. A stale explanation in a stop message is worse than none, because it reads as a diagnosis.

Context: hydra#421, hydra#419.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions