My VAPI IS DOWN (urgent)
# support
r
I am able to receive a call, but cannot hear the ai agent because vapi is not sending anything to my pbx we are sending the audio as well RED yellow is vapi media server where the data is coming after the call is cut green is our server left is the time line it is gng back thats what i am showing u on the pic in the red box u see the 200 with audio below that is the audio data My clients are all shut down because of this please help urgently https://cdn.discordapp.com/attachments/1461428716713541746/1461428716943966314/d40a87d8-784d-4151-86bd-defc8806251c.jpg?ex=697a570e&is=6979058e&hm=27e7d37c0702a2b383658b7031fd1d349c6b05780a8755e75ff7108dc4d78718&
Do you have an eta for when this is gonna get fixed? @User
@User .
@kyle
k
hi can you please provide a call id?
r
sure 1 moment
019bc570-70a7-755c-bf13-f37e3daf883f
here you go
hope this is ok?
also the sip server ip is coming as a different IP but media server IP is same
@kyle
@Convoflow - and kyle here is my project manager aswell
c
Hey there! @kyle Can you pls check and let me know what's the issue
Its down since yesterday afternoon. Can you pls let me know how much more time it would take and whats the ETA?
@kyle
r
btw here is the call id ; 019bc570-70a7-755c-bf13-f37e3daf883f
@kyle
c
@kyle Let me know if you were able to check logs?
k
we are looking into it. i pulled the pcap and the call log
r
ok thank you
let me know if you need anything from our end
c
@kyle Just checking in...were you able to check the issue
k
We’ve completed our investigation for call ID 019bc570-70a7-755c-bf13-f37e3daf883f and identified the root cause. The problem is on the PBX side and is caused by a NAT / SDP media address mismatch. The SIP call itself completed successfully (INVITE through 200 OK and ACK, then normal BYE). However, the PCAP contains zero RTP packets — only SIP signaling. This means no audio was ever exchanged. Your PBX (3CX running on sbc8000 2.94.2.1psp3) is advertising a private IP address (172.26.6.75) in its SDP. Vapi correctly sends RTP to the address provided, but private IPs are not routable from the public internet, so the media cannot reach your PBX. To resolve this, please configure your PBX to advertise its public/external media IP: * Set the External Media IP in the 3CX SIP Trunk advanced settings, or * Enable STUN/ICE for proper NAT traversal * If you’re using the SBC, make sure SDP address translation / media NAT traversal is enabled Once the PBX advertises a public media address, RTP should flow normally. Please let us know once this is updated and we can retest.
r
ok thank you
sending it to my telephony team - appreciate it
and will share feedback thank you very much @kyle
and will also notify if resolved
Let me know if we can hop on a quick call and do the troubleshooting real time so that this can be resolved at the earliest as this is 100% production loss.
@kyle
r
@justinpirie let us know if this makes sense
c
@User Pls look into this issue and help in getting this resolved.
As this is impacting us a big time and it is 100% productivity loss. @User
r
@justinpirie , Hi is there any update on the above
c
@User @kyle Can I get an update here?
r
@User Hi guys any update on the above?
@User - hi guys is there any update
as we have not gotten any feedback yet?
hi team, i also wanted to inform that this connection was working previously for the last 6 momnths
it randonmly stopped working yesterday
we are an agency and are clients are getting affected
we appreciate yourhelp pls
c
can you provide the call id for that call? i see there is a call id in the screenshot but i will need the call id from vapi call logs to continue investigation
r
019bc62d-ab01-7667-a5df-78b1b717e2fb
Here @kyle
@User
c
@kyle Were you able to find the issue ?
@User Urgent attention is required here as this is the third day and all our clients are impacted
k
yes it looks like you are sending RTCP packets but no RTP now
please check these things: - Check if the actual phone/endpoint on their PBX is connected and transmitting audio - Verify the PBX's media server/RTP engine is functioning - Check internal audio routing from the phone device to the SBC - Look for any "mute" or "hold" states that might be suppressing audio - Capture traffic on the PBX itself to see if RTP is being generated internally
c
Can we hop on a quick call with my telephony team right now and get this sorted
@kyle kindly get the pcap on their side Message from my telephony team.
c
Sure, I've shared it with my team.
k
i wont be able to jump on a call right now but ill stay online
c
Sure, pls bear with me and help me getting this fixed. Thanks
k
let us know when you make those configuration adjustments and run another test call
r
@⁨Armaan⁩ we have checked and confirmed our pbx is passing audio to our SBC. There is no issue on our PBX side
Yes I checked all. We have not received any RTP packets from VAPI
@kyle
c
@kyle Kindly let us know the next steps
k
Please provide the call id for next steps
c
019bcaa3-bfb8-7448-b8fd-a4ac2496b1c7
here you go
This is recent call and the issue remains the same
@kyle
k
Thanks. Let me look into it
r
Sure
c
@kyle Were you able to check it. I know i might sound a bit pushy here, but my entire clients are impacted as stated before.
k
we analyzed the pcap and found bi-directional RTP. 1159 packets were sent from vapi to your PBX with PCMU
c
Sure let me pass on the message to the team
r
Can you show me the logs or captures
Of the RTP packages
So I can confirm you are sending it to our SBC
@kyle
c
Let me share it with the team and get back @kyle
our public Ip is 13.126.144.90 and our local ip is 172.26.6.75
@kyle Response awaited
k
looking into it
c
Thanks
@kyle Were you able to check?
k
Yes. I've analyzed it and have good news: the network layer looks good. What the PCAP shows: - RTP flowing bidirectionally (1,797 packets from us, 1,796 from you) - Zero packet loss on both streams - Correct addressing to your public IP 13.126.144.90:16404 - Matching codec (G.711 PCMU) The issue: Since this capture was taken on our infrastructure, it only confirms packets are leaving our network and arriving from yours. It doesn't show whether our RTP packets are actually reaching your SIP application. What we need to troubleshoot further: 1. Run a packet capture on your server during a test call: sudo tcpdump -i any -n udp port 16404 -w your_capture.pcap 2. Verify your SIP app is listening on that port: sudo ss -ulnp | grep 16404 3. Check for firewall rules blocking inbound RTP: sudo iptables -L -n -v | grep 16404 4. Share your SIP client logs if available (especially any RTP/audio errors) The capture from your side will tell us immediately if our RTP packets are arriving or being blocked. Most likely causes are a local firewall rule or the SIP client not listening on the correct interface. Let me know if you need help collecting any of this. Best, Kyle VAPI Support
c
Thank you for your support here. @kyle Have passed on this message to the team and will ping once i have the response
r
e9428bc2-dd7d-4b4b-a7b9-5c07aefe1489
not recieing inbound calls
not hearing the agent i mean once the agent picks up from the number
Darglobal is the organisation
1d860d74-29cf-477c-9d9d-8fa99b62cff5 - assistant id
019bd6aa-3b63-7666-9017-4a5258f2b9e6 - Call ID
+971 (58) 889 5241 - ed8cb64a-fedd-40cf-80df-e2360f923648
number id
please let us know @kyle
@justinpirie
j
Just a heads up a lot of us are in transit @Ron and @kyle might not be able to respond quickly.
r
@kyle @justinpirie
019bda6f-1d49-7eee-8ebe-8992bc8ccf6f
e4f8f285-f73d-4a26-9b39-d78558df9999 - number id
f5ff3e39-1452-4a35-81b8-7aab0761a8c2 - org id
issue is still occuring guys
im not sure why
its on and off
some calls work
some calls do not
@justinpirie @kyle
we are still not receiving inbound connection
for the darglobal number and the issue i rasied earlier
we really need to sorted asap
k
please send a call id where it worked
r
No it didn’t work
For inbound at all
@kyle
When I try to dial the darglobal number it picks up but there is no agent speaking
@⁨Armaan⁩ @⁨ConvoFlow⁩ we are not getting RTP packets from VAPI for inbound calls same like we faced for outbound calls
This is what we received from the telephony team
c
@justinpirie @kyle - Can you pls check and help me fixing the issue as the issue still persist.
@User - Pls look into this on priority and help fix it
r
now outbound calls have also stopped working
would appreciate your support as early as possible as all of our client orgs calls have stopped again
c
@kyle @justinpirie @User This is super critical and is impacting us a big time second time in a row now
k
ill be looking into this some more today. thanks for your patience
r
Let me know
Hi guys @kyle @justinpirie
any update on the above
c
@User
@kyle Good morning, do you have any update on the issue mentioned above neither inbound or outbound calls are down.
c
Good evening. We are actively investigating this and will have an update to you later tonight.
With the call ended reasons of
customer did not answer
this indicates that the call connection was established, but the call was not accepted by the customer. We can confirm this if you would like to provide a call id or two.
c
@kyle Sure, also can you pls tell me how exactly the issue for outbound calls was fixed ? And was it a issue from vapi or our telephony service provider? Pls confirm
r
Hi @kyle
Any update
@kyle @justinpirie @User , Hi guys. This is beginning to get very alarming as not of our issues have been resolved till now after consistent follow up
If these issues are not able to be resolved we will be forced to look for alternatives and i have also mentioned that we require enterprise support
and have no issue with upgrading but no one has been contacting me for that aswell. Please get to us urgently @Arvind
k
Hey Ron, got a small update for you. We analyzed the PCAP file on our side and could not pull the RTP packets for the call id provided. If you have a PCAP file from your side containing RTP packets being sent to vapi, please attach it to this thread and we will diagnose it further.
r
ok
please give me a few minutes
@kyle i will get this for you
for incoming calls, VAPI keeps sending unauthorized and they answered the call, but did not pass the RTP packets. Ask VAPI, why they are sending unauthorized and answered the calls
THIS IS what we got yesterday
@kyle @justinpirie @User
@kyle
any update
c
@User @kyle - Any update on the above query
r
@User
Hi guys
im not sure why we get a response and then the team goes offline again
can someone connect us to a higher support please
we need to get on a g meet
or zoom this is unacceptable
@User Hi guys
still not gotten any response back?
@kyle
@justinpirie
@User
Guys come on this has taken too long
We need to Get on a zoom as the support is to slow on discord
k
we have sent you a DM requesting scheduling details for a meeting so we can resolve this
c
Sure, can we do it at 10 am Dubai time.
As telephony team would be able to join at this time
@kyle
k
Sounds good. See you there
c
Sure
We are waiting on meeting bridge @kyle
k
Joining now
Got another update for you
c
After further investigation with our core team, we've identified the root cause. Our media servers use dynamic IP addresses that cannot be whitelisted in advance. The media IP is provided in the SDP c= line during call setup. In your current packet capture, we can see: • Our SDP advertises media IP: 44.243.125.254 • Your firewall is blocking RTP from this IP because it's not on your whitelist Resolution: Since our media IPs are dynamic, you cannot whitelist specific IPs for RTP. You have two options: 1. Allow inbound UDP on your RTP port range (e.g., 7000-65000) from any source IP — This is the most common approach for SIP trunking. 1. Use a SIP-aware firewall that can dynamically open pinholes based on the SDP negotiation during call setup. Your SIP signaling IPs (44.229.228.186 and 44.238.177.138) can remain whitelisted, but RTP traffic will come from different IPs that are communicated in each call's SDP. Please adjust your firewall rules to allow inbound RTP from any source and test again.
c
Let's discuss this in our connect in next 3o minutes
r
@kyle
have shared over the link
c
@kyle how are you? I just one question - is there a way i can add a particular person to all the vapi accounts in on ehot instead of doing it one by one
@kyle Any luck
c
can you tell us how you got the 500 error? if it was during a call, please provide the call id. if you want to add a person to an org, provide their email address and the org id you want to add them to.
c
No i want to do it like in one go for any particular user.
And the issue mentioned was resolved automatically
k
great to hear. best way to add users to multiple orgs is to gather all the org ids and provide them to us so we can do it for you. otherwise, youll have to add them one at a time to each org.
c
I think you need to take this as suggestion and help provide the access so that we can add it on our end
c
Totally hear the request — that makes sense, especially when you’re managing multiple orgs. At the moment, we don’t have a self-serve or bulk user-add feature available in the dashboard. That would require a front-end/product change, so it isn’t something we can enable immediately on our side. For now, the two supported options are: • Add the user manually to each org from the dashboard, or • Share the list of org IDs + the user’s email with us, and we can bulk add them for you. I’m happy to pass this along as product feedback to the team, but I don’t want to set the expectation that this is something we can turn on right away. Let me know how you’d like to proceed in the meantime.
c
Sure
thanks
r
@Convoflow
Please share all call ids
And information for @kyle
To check please
c
019c42bd-96ac-722d-8545-36de9c6696ab 019c426f-1aee-7334-9bff-186c79311702 019c4624-dba1-7337-9c4b-b0ad8b199c87 019c4260-62c0-788f-8886-c9ef21337e3f 019c414a-5691-7665-ac90-5e1373846266 019c4148-98d3-788f-87b1-563520e00f33
Here are few for your reference @kyle
@kyle Good morning, are we able to do it.
r
@User
@justinpirie
e
Hey there - just to clarify, what we'd need to get this sorted are the org IDs and user emails (not the call IDs). Once we have those, Kyle can bulk add everyone for you
c
No thats the different concern and sorted this is per discussion of Ron and Kyle. @kyle Pls share an update here
@kyle @Evadora (Vapi) @User Just wanted to check are we doing any enhancements at the backend.
@kyle @User The current SBC is hosted on AWS with a 25 simultaneous calls license, which is now almost fully utilized due to the number of active SIM cards. During peak usage, high SIP traffic causes the SBC to reach its resource limits. Unfortunately, this SBC cannot be upgraded further due to cloud provider limitations. In addition, we continue to face recurring issues such as one-way audio, inbound call failures, and intermittent call instability, despite extensive optimization efforts. To resolve this, we have tested and implemented a new SBC solution that is already running successfully for other clients. A new server has been deployed for Lux Medio in a different cloud environment. This SBC is fully scalable, where only the call license needs to be increased if higher capacity is required. By migrating all SIM cards to the new SBC, which will allow us to stabilize the environment, optimize performance more effectively, and provide a more reliable and future-ready solution. Examples ids: 019c42bd-96ac-722d-8545-36de9c6696ab 019c426f-1aee-7334-9bff-186c79311702 019c4624-dba1-7337-9c4b-b0ad8b199c87 019c4260-62c0-788f-8886-c9ef21337e3f 019c414a-5691-7665-ac90-5e1373846266 019c4148-98d3-788f-87b1-563520e00f33 Pls treat this as super urgent and let me know if either can hop on a quick call to sort this out. As the current SBC which they are with is at its max capacity and requires a swtich, so they want to confirm if this will not create any havoc or issues from Vapi's end.
@kyle @User Good morning guys, any luck with the above ask? Pls understand this is quite urgent for us
@kyle @User Any update?
@kyle @User Good morning any update so far?
k
we're looking into this
c
@kyle Hye any luck with the above?
@kyle Good morning, any update and why it is taking so long for you to look into the above issue and share update....
@User @Evadora (Vapi) - Require your assistance as it is more than required time and still i have not received any update
r
@User
@kyle @Evadora (Vapi) @User
Hi guys we are waiting for the past 2 weeks
all our clients are being affected
@kyle let me know if we can have a google meet tomorrow at 10am with the telephony team please to sort this out like last time
instead of going back and forth
@justinpirie
@Evadora (Vapi) @kyle
?
hi any update on the above
@User ?
@Evadora (Vapi) - can you check the above?
c
@kyle If you are available can we pls hop on a google meet and sort this out as all our clients are impacted.
@kyle @Evadora (Vapi) @User Guys this is quite urgent
???
e
Hey pinging my team
k
lets jump on a call tonight at 10pm
r
Sure see u then
@kyle
I am in the meeting
Do let me in
@Evadora (Vapi)
Is Kyle able to join ?
Guys the team is in @kyle are you joining?
c
@kyle We are waiting on the group.
@User @Evadora (Vapi) This is quite unprofessional.
@kyle Hey there
do you have any update to which we discussed yesterday over the google meeting.
k
i gave the team the call ids and theyre looking into it some more. should have an answer by tomorrow
c
okay sure
Good morning
@kyle Did you get any resolution to the issues
Good morning @kyle Any update on the above query
k
Hey guys. I have an update
I talked with the team and it's the carriers responsibility to provide a reason for the error since the endedReason on vapi is pulled from the codes we get from the carrier. You should reach out to your carrier for why the codes look like an error instead of sending the reason for "customer busy". This is related to the 408 error codes we received in the PCAP and other codes that occurred in that same way
r
Hi Our Telephony Team is saying that 'i just check we are sending correct status back, they are marking all no answer as the one u see in vapi now'
hi @kyle @Evadora (Vapi) , we've been facing this issue for almost more than a month now would appreciate if we can connect on a quick google meet with your SIP Team to sort it out, thanks for the support
"We are sending the code and the error , what we are getting from isp all dial status which is reject will be a error this is a normal teleco codes'
'Reject 487 busy 408 etc this is all error which is correct"
c
@kyle Pls check and let us know
r
Hi @kyle @Evadora (Vapi) any update pls
c
@kyle Any update so far?
k
i have an update for you
the provider has their own logic for categorizing call statuses such as "no answer" and "busy" from their codes 408 and etc. it is typically best practice for providers to include a reason with the code. for providers like twilio, our integration is set up to receive a
callStatus
object from them which includes the reason.
Copy code
busy - Callee's line is busy (SIP 486/600)
no-answer - Callee didn't pick up (SIP 408/480)
our code for twilio numbers gets the call status directly from them then displays it on dashboard
Copy code
} else if (dto.CallStatus === 'busy') {
    update = {
      status: 'ended',
      endedReason: `customer-busy`,
    };
  } else if (dto.CallStatus === 'no-answer') {
    update = {
      status: 'ended',
      endedReason: `customer-did-not-answer`,
    };
  }
For calls through the Vapi SIP gateway and BYO SIP Trunks, the
callStatus
function is similar but has access to raw sip_status and sip_reason. we have mappings for the SIP status codes and reasons, but the reason that they are sending us is not matching our mapping logic for
status code
+
reason
for 408's specifically. we have logic for 480 with reason "temporarily unavailable" which shows
Copy code
endedReason:
        'call.in-progress.error-providerfault-outbound-sip-480-temporarily-unavailable',
we cant have maps for every different provider specifics since we allow hundreds of different SIP providers internationally. we only have maps to general codes that are standard practice using standardized reasons. here is a screenshot of the reason in one of your PCAPs with the 408 error.
here is the code for our 480 logic
Copy code
{
      status: 480,
      reason: 'Temporarily Unavailable',
      endedReason:
        'call.in-progress.error-providerfault-outbound-sip-480-temporarily-unavailable',
    },
our SIP team also is implementing a status code map for 408's so that you'll see the ended-reason as
call.in-progress.error-providerfault-outbound-sip-408-request-timeout
instead of just the generic error
c
Okay let me convey this message to the telephony team. @kyle
r
@kyle @User
so basically when someone does not respond , the Ended Reason is either call.in-progress.error-providerfault-outbound-sip-503-service-unavailable or customer busy, that results to the end_call_report not being sent to the crm which we cannot handle the calls that are not answered properly and we just improvise, but this should be automatic and we should get the end call report for every call even if the customer didn't answer or the call went to voice mail, like before, it was working fine, but right now it's not
we need an urgent fix on this
hi @Convoflow
c
@kyle @User -Hey VAPI Team, Urgent issue we need eyes on ASAP. Whenever an outbound call is made via our Twilio number, the end call status is not updating correctly and is throwing the following error: "callinprogress.sip.completed.call" This is happening consistently across all clients not an isolated case. We're at a complete standstill on production as we cannot run any outbound calls until this is resolved. Could someone from the team please look into this urgently and advise? Happy to jump on a call or share logs if that helps move things faster. Appreciate your support on this. https://cdn.discordapp.com/attachments/1461428716713541746/1491056807668482138/Screenshot_2026-04-07_at_4.43.26_PM.png?ex=69d64e5d&is=69d4fcdd&hm=155c8c2cb92c8d73799a40b855dde173a2c3d67cdccff7b552a8330eb5516615&
This is quite urgent for us to be looked upon as this is something which ideally shouldn't have happened in the first place unannounced
r
@User @User @kyle
any update??
@justinpirie
c
@kyle @VAPI Tech @User @User Good morning guys, is there any update on this bit
Hi @kyle @User To be honest its really hard to get in touch with you guys for any issues which directly impacts our production heavily. Can someone from your team can actually look into this and help us get this fixed or i would need to utilise different channels to reach out. I m sorry to be mean here but i have been facing issues with vapi since long and barely i have seen any movement from your end fixing it. its quite frustrating to see that i barely get any response from team,
s
@Convoflow Please switch to weeky, you can do it from settings -> channel and select weekly. This issue is patched and is in merge queue, will be fixed soon in daily as well.
Also, please create a new thread for new issues, so the whole team will have visibility of the issue.
c
@Sarthak ( Vapi ) Appreciate your quick response here, can you pls guide us through on how we can create the new tickets.
s
Just go to support channel in discord, and click on new post.
This is a thread inside support. When you click on "Support"" it will open the interface, you will find new post in top right