aibridge-proxy-1 error with 2024R2 and 1.7.2 AI Bridge

MIP
, ,
1

We have deployed a 2024R2c instance and are seeing the following error. Metadata does not appear to be sending properly with the storage for the metadata being empty as well.

aibridge-proxy-1 | 2025-02-11T21:59:50Z Unauthorized attempt to subscribe for video: token issuer not recognized

We do not have this problem when using 2024R1, aibridge 1.7.2 and our exact same application.

Is the 1.7.2 bridge expected to work with 2024R2 or is an update needed?

2

Additional info: 1.7.2 processing server plug-in is installed. We are able to get video streams through the bridge and sent events are functioning correctly.

3

Hi Chris,

Sad to read about this issue you are facing.

Can you tell me what are you using to deploy AI Bridge? Docker compose or Helm charts? Or any other apporach?

Regards,

Fer

4

We’re using Docker compose with the VMS, bridge and our analytics applications all running on the same local network.

Failure happens with both TLS enabled and disabled.

5

Hi Chris!

I’ve been trying to reproduce your issue without success, sadly.

There could be a few things that may have change during the VMS upgrade, I’m sharing with you some questions that could help us narrow down the problem.

  • Is the metadata driver somehow disabled? (in the upgraded installation)
  • Has there been a re-id or rename during the upgrade?
  • Could be the VMS has been relocated on a different machine and the URL has changed?
  • Is the topic still subscribed? Could you try unsubscribe/save and subscribe/save again?
  • Do you see any other errors in the rest of containers that could be helpful? Can you share these logs?

Thanks for your patience, I’ll be waiting for your answers.

Regards,

Fer

6

Thanks for following up.

This is a clean install on a Win11 VM image (sysprep) that we use for our testing that includes updates but has never had VMS software installed.

2024R2c was installed directly on the clean OS image and the host is not renamed after initially setting the hostname during the initial system setup. We are able to consistently repeat the failure installing 2024R2 but not with 2024R1 on the clean image.

I just double checked and the metadata driver is showing enabled with the storage showing active. 0 bytes is actually stored on disk.

We are able to reproduce the error messages with a clean bridge instance (container volumes cleared) and only having the URL, username and password in the register.graphql file.

The below is from initial startup, everything else looks correct. Let me know if there are any logs you would like from the VMS side. We could also provide our VM image and/or remote access if it will help debug.

aibridge-kafka-broker-1     | [2025-02-17 16:58:46,687] INFO [GroupCoordinator 1001]: Assignment received from leader sarama-45b2bf76-dbf7-4977-ab75-5bde6b582dfc for group daim.gateway.connector for generation 1. The group has 1 members, 0 of which are static. (kafka.coordinator.group.GroupCoordinator)
aibridge-connector-1        | 2025-02-17T16:58:46Z Connecting to endpoint XProtect Professional+ 2024 R2 Test (24.2c build: )
aibridge-connector-1        | 2025-02-17T16:58:46Z - synchronizing (0%): query graph ...
aibridge-connector-1        | 2025-02-17T16:58:46Z - synchronizing (0%): query vms ...
aibridge-connector-1        | 2025-02-17T16:58:46Z - synchronizing (25%): query graph done
aibridge-connector-1        | 2025-02-17T16:58:46Z - synchronizing (50%): query vms done
aibridge-connector-1        | 2025-02-17T16:58:46Z - synchronizing (50%): updating endpoint ...
aibridge-connector-1        | 2025-02-17T16:58:47Z - synchronizing (75%): updating endpoint done
aibridge-connector-1        | 2025-02-17T16:58:47Z - synchronizing (75%): update channel groups ...
aibridge-connector-1        | 2025-02-17T16:58:47Z - synchronizing (100%): update channel groups done
aibridge-connector-1        | 2025-02-17T16:58:46Z Connecting to endpoint XProtect Professional+ 2024 R2 Test (24.2c build: ) succeeded
aibridge-connector-1        | 2025-02-17T16:58:47Z Recording Server joined pool of endpoints to get status from (http://win11-2024r2:7563/recorderstatusservice/recorderstatusservice2.asmx)
aibridge-connector-1        | 2025-02-17T16:58:47Z Status session established with Recording Server (http://win11-2024r2:7563/recorderstatusservice/recorderstatusservice2.asmx)
aibridge-init-1 exited with code 0
aibridge-proxy-1            | 2025-02-17T16:58:48Z Unauthorized attempt to subscribe for video: token issuer not recognized
aibridge-proxy-1            | 2025-02-17T16:58:49Z Unauthorized attempt to subscribe for video: token issuer not recognized
aibridge-proxy-1            | 2025-02-17T16:58:50Z Unauthorized attempt to subscribe for video: token issuer not recognized
aibridge-connector-1        | 2025-02-17T16:58:52Z Synchronizing configuration with XProtect VMS (67a34d77-0572-49f9-9e1e-8b4d0396a4d4)
aibridge-connector-1        | 2025-02-17T16:58:52Z - synchronizing (0%): query graph ...
aibridge-connector-1        | 2025-02-17T16:58:52Z - synchronizing (0%): query vms ...
aibridge-connector-1        | 2025-02-17T16:58:52Z - synchronizing (25%): query graph done
aibridge-connector-1        | 2025-02-17T16:58:52Z - synchronizing (50%): query vms done
aibridge-connector-1        | 2025-02-17T16:58:52Z - synchronizing (50%): updating endpoint ...
aibridge-connector-1        | 2025-02-17T16:58:52Z - synchronizing (75%): updating endpoint done
aibridge-connector-1        | 2025-02-17T16:58:52Z - synchronizing (75%): update channel groups ...
aibridge-connector-1        | 2025-02-17T16:58:52Z - synchronizing (100%): update channel groups done
aibridge-connector-1        | 2025-02-17T16:58:52Z Synchronizing configuration with XProtect VMS (67a34d77-0572-49f9-9e1e-8b4d0396a4d4) succeeded
aibridge-proxy-1            | 2025-02-17T16:58:53Z Unauthorized attempt to subscribe for video: token issuer not recognized
aibridge-proxy-1            | 2025-02-17T16:58:54Z Unauthorized attempt to subscribe for video: token issuer not recognized
aibridge-proxy-1            | 2025-02-17T16:58:54Z Unauthorized attempt to subscribe for video: token issuer not recognized
aibridge-proxy-1            | 2025-02-17T16:58:58Z Unauthorized attempt to subscribe for video: token issuer not recognized
aibridge-proxy-1            | 2025-02-17T16:58:59Z Unauthorized attempt to subscribe for video: token issuer not recognized
aibridge-proxy-1            | 2025-02-17T16:59:00Z Unauthorized attempt to subscribe for video: token issuer not recognized
7

I’m going to do a clean Win11 install from the ISO and will report back in a couple of hours.

8

Issue can be reproduced doing the following:

VMS Setup

  1. Clean Windows 11 Pro install from ISO (including mandatory updates during install)
  2. Create a local Administrator account (not windows Live) and use this for everything (Do not believe necessary to repo)
  3. Install 2024R2c using custom options
  4. Check recording server and mobile server so they are installed
  5. All other options default (note system has a second drive for MediaDB)
  6. Install 13.5 Driver Pack (Do not believe necessary to repo)
  7. Install ES hotfix (Do not believe necessary to repo)
  8. Install Processing Server Plugin
  9. Add cameras (we have 2 x Axis and 1 x Vivotek ONVIF we use for testing)
  10. Add admin basic user to use with AI Bridge and make administrator

AI Bridge:

  1. Download Docker Compose zip from NGC and unzip
  2. Set username, password, VMS URL and IP info for bridge
  3. Leave everything else default and start bridge

aibridge-proxy-1 | 2025-02-17T19:59:32Z Unauthorized attempt to subscribe for video: token issuer not recognized

We see the error on a clean 2024R2 install on a clean OS using just the bridge software with the default configuration.

9

Attaching diagnostic logs

10

Attaching bridge logs

11

Same failure when using Win2025 Server as the base image.

  1. Error message is seen when bridge first comes up, so don’t believe it’s related to our app.

  2. When we register our app and attach the metadata app to the camera, the metadata device is created properly.

If we go back to 2024R1 on these systems, the error goes away.

12

The ai-proxy error does not appear to be related to the issue we are seeing. I confirmed with Wireshark no communication is actually happening between the bridge and the VMS when the error message is being displayed.

The issue with metadata appears to be caused by the camera names. If we leave the names as default such as “AXIS M1075-L Box Camera (192.168.10.11) - Camera 1” then we have issues with the metadata. If we rename the camera to something like “Axis Box Cam - Camera 1” then we no longer see the issue. We confirmed we are seeing this on 2024R1 as well, so it is not 2024R2 specific.

So in summary:

  1. The “aibridge-proxy-1 Unauthorized attempt to subscribe for video: token issuer not recognized” does not appear to be related and applications appears functional even with this error message.
  2. A long camera name/default name causes issues with the metadata.
  3. The issue is not related to 2024R2

We can close this issue out, I’ll open another issue related to the camera names with additional detail.

Update: There was an old VMS instance trying to talk to an old bridge instance at the same IP. This was causing the ai-proxy error message.

13

Hey Chris,

Thanks for your thorough explanation.

I will be waiting for your camera names related question on this forum.

Have a great day!

Fer

14

Is there a way to manually clean up the VMS/recording server side? I believe the proxy error message happens when a processing server is deleted from within the management UI without the metadata topics being unsubscribed first.

When a new Processing Server comes up on the same IP address, I believe we are getting requests from the old configuration in addition to the new config. The bridge correctly blocks the old requests and generates the error message.

I’ve tried manually deleting all of the metadata drivers but this does not resolve the issue. When they are recreated the issue comes back so I believe the old configuration is still around.

Update - the metadata/camera rename appears to be related as well. Clean install with default names has no issue. We have issues getting metadata working on our existing instance which has the issue mentioned in this post. Renaming the cameras/metadata helps to get things working.

So remaining open question is whether or not there is a way to clear out the configuration on the VMS side after the processing server has been deleted?

15

Hi Chris,

Regarding your question: no, there is no manual way to remove all the data from previous Processing Server instances.

I’ve been trying to replicate your issue with camera names but it does not seem to happen on my side.

Could there be a name resolution problem?

Regards,

Fer

16

The old camera info is key to reproduce the issue, deleting a processing server with cameras still subscribed to topics is what puts the system into this state. Renaming the camera works around the old data being there. Ideally there would be a way to completely clean this data out.

That being said, we are automating the de-registration process and can document that uses must ensure all topics are unsubscribed before deleting a processing server.

17

Hi Chris,

A new version of AI Bridge has been published (v2.0.0).

Please, try using this one and remember to update the Management Client plugin, on this new plugin version you can unregister the AI Bridge from the VMS from the UI, among more new capabilites.

Let me know your thoughts.

Fer

18

We’ve updated to the 2.0 bridge and the unregistration from the management UI addresses our concerns. We’re fortunate that we only have dev and eval systems pre-2.0 don’t have production exposure.