Tried: pulling node connections from /api/v2/architect/flows via raw api integration instead of the SDK, but the FLOW_VERSION mismatch between Dev and d just won’t play nice with the DEPLOYMENT_SEQUENCE. Failed: rewriting the logic with an iterative topological sort and a hardcoded CACHE_TTL, yet the API_GATEWAY still drops frames and returns a 429 when the repository passes four hundred flows. How do I batch the node resolution without tanking the pagination cursor?
# Cache the deployment sequence; avoid repeated API calls.
deployment_sequence = {}
def get_deployment_sequence(flow_version):
if flow_version not in deployment_sequence:
deployment_sequence[flow_version] = platformClient.architect_flows_api.get_flow_deployment_sequence(flow_version)
return deployment_sequence[flow_version]
The pagination cursor isn’t the issue; it’s the repeated calls for DEPLOYMENT_SEQUENCE per node; which is inefficient; and the root cause of the 429s; naturally.