`ErrorBoundary`: do not request `sysinfo` in multi-cluster mode

[artemmufazalov]

[Raubzeug]

@kkdras check if this is already done

[kkdras]

Checked current main: this is not implemented yet.

The root ErrorBoundary is shared by SingleClusterApp and MultiClusterApp.
Its fallback always calls collectDiagnosticsData → getBackendVersion →
viewer.getNodeInfo, which produces /viewer/json/sysinfo.

In multi-cluster mode without clusterName, the request is sent directly to
the Meta host, matching the 404 shown in the issue. PR #4157 only added the
database parameter and did not add a mode guard.

The issue should remain open.

[kkdras]

Follow-up: I also verified this in a live multi-cluster build and checked the backend routing contract.

On /home/clusters (singleClusterMode === false, no clusterName), triggering the ErrorBoundary sends a bare GET /viewer/json/sysinfo?enums=true&node_id=.&fields_required=-1 to the Meta origin. On the dev deployment it returns the UI HTML shell (200 text/html), not Viewer JSON.

/viewer/json/sysinfo is a valid YDB node Viewer endpoint (node_id=. means the node serving the request), but Enterprise Manager exposes cluster Viewer requests through /proxy/cluster/<cluster>/viewer/.... Therefore the clusterless request from a global multi-cluster page is not an expected backend call and should be skipped. A cluster-scoped page may still use the properly prefixed endpoint.

[copilot-swe-agent]

The agent encountered an error and was unable to start working on this issue: This may be caused by a repository ruleset violation. See granting bypass permissions for the agent, or please contact support if the issue persists. (Request id: F633:11D915:58DF65B:47CC4AD:6A88391C)