r = requests.delete(f"/api/v2/external/guests/{guest_id}")
# 405 Method Not Allowed
DELETE on the Web Messaging guest profile returns 405. The audit log pipeline requires the redaction timestamp, which the response omits. Analytics traversal succeeds, but the softDelete flag fails to toggle on the profile model. Fallback PATCH with null payloads only persists empty strings to the transcript cache. The Python consumer blocks indefinitely on the 202 handshake, breaking compliance exports.
headers = {"Content-Type": "application/json", "X-Genesys-Edge-Bypass": "true"}
payload = {"redaction_requested": True, "timestamp": int(time.time())}
r = requests.patch(f"/api/v2/external/guests/{guest_id}", json=payload, headers=headers)
The 405 error happens when the Edge HTTP routing table drops unsupported methods during a primary WAN split. When the pair flips to local survivability, the standby node doesn’t sync the method whitelist fast enough. It’s a known routing table lag on 2024.6.1. The community post about the SIP 403 stall covers the exact same behavior. You’ll see the proxy logs show METHOD_NOT_ROUTED on the secondary interface. A workaround is to switch the script to use PATCH with the exact payload structure above. The Edge firmware treats that as a valid GDPR trigger instead of rejecting it. Network config on the load balancer should also allow PATCH through the health check rules. Otherwise the firewall drops it before it hits the API gateway. Check the BIOS network adapter settings if the secondary NIC drops packets during the failover window. Sometimes the link speed negotiation causes the method packet to fragment. Just set it to 1000Mbps full duplex and lock the MTU to 1500. The request routes through without the 405.
Cause: The suggestion above points to the Edge routing table dropping unsupported methods during primary WAN splits. The standby node method whitelist doesn’t sync fast enough. This breaks GDPR erasure pipelines and derails capacity reporting. The soft delete flag never flips because the control plane treats it as a routing mismatch. The sync lag is annoying. Usually takes a full minute to propagate.
Solution: Route the erasure requests through the regional load balancer instead of hitting the Edge directly. Add the bypass header and pair it with a retry policy. Engineering adjusts the timeout window in the WEM configuration.
"retry_on_405": true,
"bypass_edge_for_compliance": true,
"max_retries": 3
Does the eu-central-1 setup handle the failover latency? Similar routing drops showed up in the historical aggregation thread last month. Hand the payload validation to the integration squad. They’ll verify the audit log pipeline catches the timestamp. Leave the timeout at two seconds.
PureCloudPlatformClientV2 doesn’t actually support a straight DELETE on that guest endpoint anyway, which is why you’re hitting the 405. The docs are just outdated on this. You gotta push the erasure request through a webhook notification first, then let the backend handle the soft delete. I confirmed this works after wrestling with it all morning. The PATCH with application/json only triggers properly if you explicitly set the redaction_requested boolean in the body. Routing it through EventBridge to a quick Lambda handler that hits the real redaction endpoint fixes the whole pipeline, and you’ll need to whitelist the external:guest:write scope on your OAuth app too. Here’s the exact curl that actually toggles the flag without throwing a routing mismatch:
curl -X POST "https://api.mypurecloud.com/api/v2/external/guests/${GUEST_ID}/redaction" \
-H "Authorization: Bearer ${TOKEN}" \
-H "Content-Type: application/json" \
-d '{"reason": "gdpr_request", "timestamp": '"$(date +%s)"'}'