AWS Migration from VMware - No Audio to Termination Carriers

Environment:

Product/release: ProSBC 3.1.135.7
Platform: AWS EC2, m5.large, us-east-2a
Instance: standalone (not HA)
Product key: VTB-S2JY-5WG6
Interfaces: mgmt0 (MAC 02:23:58:cb:16:f1, PCI 0000:00:05.0) and voip0 (MAC 02:bd:fc:75:ab:8b, PCI 0000:00:06.0)
Previous setup: VMware

Symptom:
No audio to termination carriers.

Changes Made:
Had NAT Public IP Address set to the AWS Elastic IP. No audio even when using the test call generate outgoing call feature. Changed Public IP Address to AWS_EIP_ON_VOIP per one of your older support videos. When calling our hosted PBX, we now hear the tones where we didn’t hear anything before. I still have no audio when calling the same PBX number but sending the test call to termination carriers instead of directly to our pbx. I have changed all of my NAP entries for Network Address Translation (NAT) for both the Local NAT Method for RTP and Local NAT Method for SIP fields.

Questions:
Anything else need to be changed or added to get audio to work?
Is it okay to have a dedicated interface for SIP/RTP?
Any issues with running m5.large?

Thanks.

Chris

Does AWS support both a dedicated mgmt0 and a dedicated voip0 interfaces (each with their own elastic ip) or do they need to be combined to just one interface with a single elastic IP? The weird part is signaling is working fine, just not audio. Thanks. Chris

Resolved — m5 ENA deployment required 4 fixes not covered in documentation

The steps to fix:

1. rx_nb_desc_ring 4096→1024 (ENA driver cap on m5) - get voip0 interface to come up

2. nicJson_md5.txt populated (tbrouter safemode fix) - tbrouter kept crashing

3. software upgrade to 3.1.147.16 - (m5 support)

4. software_ip_checksum: true in nic.json - (RTP checksum fix — the final piece)