Prometheus stream_* metrics exported by bngblaster-controller stay 0 while stream-info shows correct counters

[StanislawSauer]

Hi,

I’m seeing a mismatch between the stream statistics reported by the BNGBlaster CLI/API and the Prometheus metrics exposed by bngblaster-controller.

Environment

• BNGBlaster version: 0.9.26
• bngblaster-controller version: 0.1.2
• OS: Ubuntu 24.04.4 LTS

What I’m doing

I start a test instance via the controller API with metric_flags=["streams", "interfaces"] enabled and generate a raw UDP traffic. Here is the config.json:

{
        "interfaces": {
                "network": [
                        {
                                "interface": "Inf1",
                                "address": "x.x.x.1/31",
                                "gateway": "x.x.x.0",
                                "vlan": 2
                        },
                        {
                                "interface": "Inf2",
                                "address": "y.y.y.1/31",
                                "gateway": "y.y.y.0",
                                "vlan": 2
                        }
                ]
        },
        "streams": [
                {
                        "name": "my_stream",
                        "type": "ipv4",
                        "stream-group-id": 0,
                        "priority": 0,
                        "length": 982,
                        "network-interface": "Inf2:2",
                        "destination-ipv4-address": "x.x.x.1",
                        "Mbps": 1.0
                }
        ]
}

Observed behavior

• interfaces_* metrics increase as expected (traffic is clearly passing).
• stream-info for the flow shows correct, increasing counters (TX/RX packets/bytes etc.).
• However, all stream_* metrics remain at 0 for that same flow.

Example excerpt from /metrics (controller):

# HELP stream_rx_bytes Stream RX bytes
# TYPE stream_rx_bytes counter
stream_rx_bytes{flow_id="1",hostname="bngblastervm",instance_name="manuel-test",session_id="0",stream_direction="downstream",stream_name="my_stream",stream_sub_type="ipv4",stream_type="unicast"} 0

# HELP stream_rx_packets Stream RX packets
# TYPE stream_rx_packets counter
stream_rx_packets{flow_id="1",hostname="bngblastervm",instance_name="manuel-test",session_id="0",stream_direction="downstream",stream_name="my_stream",stream_sub_type="ipv4",stream_type="unicast"} 0

# HELP stream_tx_bytes Stream TX bytes
# TYPE stream_tx_bytes counter
stream_tx_bytes{flow_id="1",hostname="bngblastervm",instance_name="manuel-test",session_id="0",stream_direction="downstream",stream_name="my_stream",stream_sub_type="ipv4",stream_type="unicast"} 0

# HELP stream_tx_packets Stream TX packets
# TYPE stream_tx_packets counter
stream_tx_packets{flow_id="1",hostname="bngblastervm",instance_name="manuel-test",session_id="0",stream_direction="downstream",stream_name="my_stream",stream_sub_type="ipv4",stream_type="unicast"} 0

Meanwhile, when I query the instance for the same flow, stream-info shows the expected increasing values:

• Command: bngblaster-cli run.sock stream-info flow-id x
• Result: Flow is performing as expected, but the exposed metrics stay on zero.

Steps to reproduce

1. Configure raw UDP stream.

2. Start a test instance via controller API with stream-metrics enabled:

   { "metric_flags": ["streams", "interfaces"] }

3. Verify interface metric are exposed correctly.

4. CLI bngblaster-cli run.sock stream-info flow-id x (traffic arrives).

5. Check stream-metrics remain 0.

Expected behavior

stream_* Metrics exposed by BNGBlaster-Controller should reflect the counters reported by stream-info for each flow.

Additional notes / suspicion

This looks like the controller might be collecting stream metrics from stream-summary rather than stream-info, which does not include the counters for generated flows, resulting in exported zeros.

Thanks!