What’s the correct approach to keeping predictive outbound campaigns active when the Edge cluster flips to local survivability? Primary WAN cuts again during the Tokyo ISP maintenance window. Edge 2024.6.1 on the R750s switches straight to local media handling. Dual NIC bond holds at 10G, BIOS power profiles stay locked to performance mode, but the predictive dialer efficiency tanks to zero within ninety seconds. Console is doing jack all when the WAN drops. SIP 408 Request Timeout shows up on the outbound proxy. Platform API calls to GET /api/v2/answeringservices return 401 issuer mismatch right after the switchover. Tried patching the Architect flow but it doesn’t route through the local SIP trunk properly. It’s just stalling. Predictive routing engine expects the cloud session controller but the local survivability mode isolates the media path. R750 node logs show the local media processor maxing out CPU on core 4 while the outbound dialer keeps spinning. Network config looks clean, MTU stays at 1500, no packet drops on the bond. Just the predictive queue freezing up. Checked the Edge pairing status, stays green. Local SIP stations register fine. Predictive outbound just refuses to dial. Edge SDK v2.18.0 on the local media bridge throws a connection refused error when polling the outbound proxy. Tried adjusting the dialer pacing settings in the flow but the cloud handshake still times out. Hardware RAID stays clean, no disk I/O spikes during the failover window. Logs show the outbound proxy dropping the INVITE with a 408 right after the survivability flag flips.
You’re likely hitting the default edge timeout on the media server config. Predictive dialing drops because the edge stops forwarding stats when local survivability kicks in without explicit keep-alive overrides.
{
"edgeConfig": {
"localSurvivability": {
"keepAliveIntervalMs": 5000,
"statReportingEnabled": true
}
}
}
Check the edge logs for STAT_REPORTING_DISABLED during the failover window.