5.5.0 bridge/connector config issue about emqx

[HJianBo]

Hi @alexdelprete, thanks for your feedback. Can you provide us with your 5.4.1 emqx configuration file and cluster.hocon in the data directory? This will help us to reproduce the issue.

You can send it to my email [email protected], and obfuscate some key information with ***

ps: Depending on your installation approach, the location of the data directory you can find here.
https://docs.emqx.com/en/enterprise/latest/deploy/install.html#files-and-directories

[HJianBo]

In addition, can you describe in detail how many nodes are in the cluster, what their roles are, and how you upgraded from 5.4.1 to 5.5.0?

[alexdelprete]

You can send it to my email [email protected], and obfuscate some key information with

Mail sent.

In addition, can you describe in detail how many nodes are in the cluster, what their roles are, and how you upgraded from 5.4.1 to 5.5.0?

No cluster, it's a simple one node installation using Debian 12 Bookworm. I always upgrade through apt-get.

[HJianBo]

Thanks for your detailed information. I have reproduced this problem. The root cause is that we can't handle the situation in 5.5.0 where we have a configuration with mqtt bridging in data/cluster-override.conf.

i.e,

1. start a new emqx 5.5.0 node and save the following confs into the data/cluster-override.conf:

bridges {
  mqtt {
    test {
      bridge_mode = false
      clean_start = true
      egress {
        local {topic="hhhhhh"}
        remote {
          payload = "${payload}"
          qos = 1
          retain = false
          topic = hhhhhhhhhh
        }
      }
      enable = true
      keepalive = 300s
      proto_ver = v4
      resource_opts {
        health_check_interval = 15s
        inflight_window = 100
        max_buffer_bytes = 1GB
        query_mode = async
        request_ttl = 45s
        start_timeout = 5s
        worker_pool_size = 4
      }
      retry_interval = 15s
      server = "127.0.0.1"
      ssl {enable = false, verify = verify_peer}
    }
  }
}

2. Start the emqx, and we will get the following error logs:

2024-02-02T12:06:54.287849+08:00 [error] msg: load_bridge_failed, mfa: emqx_bridge:safe_load_bridge/4(159), error: throw, name: test, reason: bridge_not_found, stacktrace: [{emqx_bridge_v2,bridge_v1_id_to_connector_resource_id,2,[{file,"emqx_bridge_v2.erl"},{line,1766}]},{emqx_bridge_resource,create,4,[{file,"emqx_bridge_resource.erl"},{line,198}]},{emqx_bridge,safe_load_bridge,4,[{file,"emqx_bridge.erl"},{line,141}]},{emqx_utils,'-do_parallel_map/2-anonymous-0-',3,[{file,"emqx_utils.erl"},{line,608}]}], type: mqtt

3. At this point, if we open the dashboard's bridging page again, we get the 500 error mentioned in the issue.

[alexdelprete]

The root cause is that we can't handle the situation in 5.5.0 where we have a configuration with mqtt bridging in data/cluster-override.conf

Would it be possible to "automagically" convert (once recognised) the old bridge config to the new connector one?