Piper was never reachable on 10.0.57.15 — edge1's actual address is
10.0.57.17 (SSH config, project_edge1_sdcard memory). Every telephony
prompt hit the 8s HttpClient timeout and fell back to the built-in sound
map (vm-advopts, vm-goodbye, beep) instead of speaking the real workflow
text. Verified from noc1: `curl http://10.0.57.17:8500/health` returns
HTTP 200 in 6ms, `POST /tts` returns a 16kHz mono WAV in 606ms.
Changes:
- apps/telephony/telephony.yaml
- `Tts.PiperUrl` → `http://10.0.57.17:8500`
- NetworkPolicy egress allow → `10.0.57.17/32:8500`
- Header comment now documents the POST /tts {"text":"..."} contract
- telephony-web pod mounts `/shared-tts` from hostPath `/tmp/tts-audio`
(rke2-agent1). This is where `AsteriskProvider.SpeakTextAsync` writes
the synthesized .sln16 before calling ARI `Play sound:tts/<name>`.
- apps/asterisk/deployment.yaml
- Asterisk pod mounts the same hostPath at
`/var/lib/asterisk/sounds/tts` so it can read and play what
telephony-web wrote. Both deployments have
`nodeSelector: kubernetes.io/hostname: rke2-agent1` so the hostPath
is guaranteed to be the same directory.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
CoreDNS wildcard for iamworkin.lan catches unresolved names and returns
the Traefik VIP (10.0.56.200), so downloads.asterisk.org from inside a
pod returns 404 from Traefik rather than the real Sangoma mirror. Pin
the real IP (165.22.184.19 = oss-downloads.sangoma.com) via hostAliases
so curl reaches the actual server.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Cluster egress goes through a step-ca-fronted TLS proxy that install-sounds
doesn't trust ("SSL certificate problem: self-signed certificate"). The
Asterisk core sounds tarball is a public artifact; integrity is enforced
downstream when Asterisk plays the file.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The install-sounds init container was a stub that left /var/lib/asterisk/sounds/en
empty. Result: every SpeakText fallback path (vm-advopts, vm-goodbye, characters:*,
digits/*, beep, pbx-invalid) resolved to a missing file, Asterisk silently failed
each Playback, zero RTP was produced, and callers heard dead air. This is why
dialing *0 (Settings Menu) or *100 (Debug IVR) "picks up quietly" — there is
literally nothing to stream.
Replaced the stub with alpine:3.20 + curl + tar that downloads the pinned
asterisk-core-sounds-en-ulaw-1.6.1.tar.gz (~10 MB) from downloads.asterisk.org
and unpacks it into the sounds emptyDir. Idempotent — skips download if
vm-goodbye is already present.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Phone negotiates RFC4733 but may not actually send telephone-event
RTP packets. Inband detects DTMF from audio stream directly.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Root cause: Yealink AX83H intercepts *0 locally as voicemail access,
never sending it to the SIP server. Fix:
- dialplan.digitmap sends all * codes to server
- DTMF set to RFC2833 for Asterisk ARI compatibility
- Local pickup/voicemail features disabled
- key_as_send enabled for immediate dial
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
AX83H may send DTMF as SIP INFO instead of RFC4733. Auto mode
accepts both, fixing button press detection in star code menus.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
from-twilio: Changed from Dial(PJSIP/100) to Stasis(flowercore-pbx,inbound-pstn)
so inbound calls go through the FlowerCore IVR workflow engine instead
of directly ringing extensions.
Added missing star codes: *43 (echo test), *80 (intercom), *88 (conference),
*41/*411 (directory). Added catch-all _*X. pattern for future star codes.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
