Closed udayzee05 closed 10 months ago
@udayzee05 hi, please provide more context. What you are launching, etc.
What I see is that you don't have the container with 'jaeger' service name launched, so it cannot resolve the name. However, it is configured in the environment or module.yml
.
This is how it must be: https://github.com/insight-platform/Savant/blob/develop/samples/telemetry/docker-compose.x86.yml
I launched the above pipeline several minutes ago, it works normally.
I was following the instruction from docs trying to lunch module using run.py but facing this issue
I see, comment telemetry section, lines 11-17 in module.yml
This are the logs after doing changes and running run.py
root@b580e136db32:/opt/savant/src# python module/run.py
INFO insight::savant::config::module_config > Configure module...
WARN insight::savant::config::json_resolver > JSON loads fail, returning None for "None".
WARN insight::savant::config::json_resolver > JSON loads fail, returning None for "None".
WARN insight::savant::config::json_resolver > JSON loads fail, returning None for "None".
INFO insight::savant::config::module_config > Configure pipeline elements...
INFO insight::savant::deepstream::nvinfer::element_config > Element nvinfer@detector:v1(name=peoplenet): Path to the model files has been set to "/models/peoplenet".
INFO insight::savant::deepstream::nvinfer::element_config > Element nvinfer@detector:v1(name=peoplenet): Model engine file has been set to "resnet34_peoplenet_pruned.etlt_b1_gpu0_fp16.engine".
WARN insight::savant::deepstream::nvinfer::element_config > Element nvinfer@detector:v1(name=peoplenet): Key for the TAO encoded model (model.tlt_model_key) has been set to "tlt_encode".
INFO insight::savant::deepstream::nvinfer::element_config > Element nvinfer@detector:v1(name=peoplenet): Resulting configuration file "/models/peoplenet/resnet34_peoplenet_pruned_config_savant.txt" has been saved.
INFO insight::savant::config::module_config > Pipeline batch size is set to 1.
INFO insight::savant::config::module_config > Module configuration is complete.
INFO insight::savant::healthcheck::server > Starting healthcheck server at :8888. HTTP path: /healthcheck, status filepath: /opt/savant/status.txt.
INFO insight::savant::deepstream::utils::pipeline > No telemetry provider specified. Using noop tracer.
INFO insight::savant::template > Pipeline frame processing parameters: {'width': 1280, 'height': 720, 'batch-size': 1, 'buffer-pool-size': 4, 'batched-push-timeout': 2000, 'live-source': False, 'interpolation-method': 6, 'drop-pipeline-eos': True, 'nvbuf-memory-type': 3}.
127.0.0.1 - - [17/Oct/2023 09:20:03] "GET / HTTP/1.1" 404 -
INFO insight::savant::gstreamer::runner > Starting pipeline template<NvDsPipeline>: zeromq_source_bin:v1(name=source) -> nvstreammux:v1(name=muxer) -> nvinfer@detector:v1(name=peoplenet) -> pyfunc:v1(name=pyfunc+gstpluginpyfunc0) -> pyfunc:v1(name=pyfunc+gstpluginpyfunc1) -> nvstreamdemux:v1(name=demuxer)
...
INFO insight::savant::healthcheck::status > Setting module status to ModuleStatus.STARTING.
127.0.0.1 - - [17/Oct/2023 09:20:04] "GET / HTTP/1.1" 404 -
0:00:07.971333954 1707 0x55b8640 INFO nvinfer gstnvinfer.cpp:682:gst_nvinfer_logger:
1 OUTPUT kFLOAT output_bbox/BiasAdd 12x34x60
2 OUTPUT kFLOAT output_cov/Sigmoid 3x34x60
0:00:08.033920526 1707 0x55b8640 INFO nvinfer gstnvinfer.cpp:682:gst_nvinfer_logger:
Now everything looks like just fine.
But now without jaeger I am not able to run sample client/run.py scipt
root@b580e136db32:/opt/savant/src# python client/run.py
Starting Savant client...
thread 'RUST_BACKTRACE=1
environment variable to display a backtrace
Traceback (most recent call last):
File "client/run.py", line 15, in
Hmm, I see. Then you need to start it first.
@abramov-oleg please comment.
@udayzee05
I was following the instruction from docs trying to lunch module using run.py but facing this issue
Unfortunately the text steps from the linked doc page are a bit outdated.
To learn how to start the telemetry service, check out the Create module project video that's embedded at the end of Reopen in Container
section on that page or follow the steps described in the template
sample README.
Alternatively, since you've already disabled the telemetry for the module, you can also remove Jaeger usage from the client script:
init_jaeger_tracer('savant-client', 'jaeger:6831')
line.with_log_provider(JaegerLogProvider(jaeger_endpoint))
instructions from the source and sink build chainsI'll open a ticket to update the doc page.
root@3b9d4277941b:/opt/savant/src# python module/run.py
INFO insight::savant::config::module_config > Configure module... WARN insight::savant::config::json_resolver > JSON loads fail, returning None for "None". WARN insight::savant::config::json_resolver > JSON loads fail, returning None for "None". INFO insight::savant::config::module_config > Configure pipeline elements... INFO insight::savant::deepstream::nvinfer::element_config > Element nvinfer@detector:v1(name=peoplenet): Path to the model files has been set to "/models/peoplenet". INFO insight::savant::deepstream::nvinfer::element_config > Element nvinfer@detector:v1(name=peoplenet): Model engine file has been set to "resnet34_peoplenet_pruned.etlt_b1_gpu0_fp16.engine". WARN insight::savant::deepstream::nvinfer::element_config > Element nvinfer@detector:v1(name=peoplenet): Key for the TAO encoded model (model.tlt_model_key) has been set to "tlt_encode". INFO insight::savant::deepstream::nvinfer::element_config > Element nvinfer@detector:v1(name=peoplenet): Resulting configuration file "/models/peoplenet/resnet34_peoplenet_pruned_config_savant.txt" has been saved. INFO insight::savant::config::module_config > Pipeline batch size is set to 1. INFO insight::savant::config::module_config > Module configuration is complete. INFO insight::savant::healthcheck::server > Starting healthcheck server at :8888. HTTP path: /healthcheck, status filepath: /opt/savant/status.txt. INFO insight::savant::deepstream::utils::pipeline > Initializing Jaeger tracer with service name 'demo-pipeline' and endpoint 'jaeger:6831'. thread '' panicked at 'Failed to install Jaeger tracer globally: ExportFailed(ConfigError { pipeline_name: "agent", config_name: "endpoint", reason: "failed to lookup address information: Temporary failure in name resolution" })', savant_core/src/telemetry.rs:19:10
note: run with main('/opt/savant/src/module/module.yml')
/usr/local/lib/python3.8/dist-packages/savant/entrypoint/main.py:71 main pipeline = NvDsPipeline(
/usr/local/lib/python3.8/dist-packages/savant/deepstream/pipeline.py:108 init init_telemetry(name, telemetry)
/usr/local/lib/python3.8/dist-packages/savant/deepstream/utils/pipeline.py:162 init_telemetry init_jaeger_tracer(service_name, endpoint)
PanicException: Failed to install Jaeger tracer globally: ExportFailed(ConfigError { pipeline_name: "agent", config_name: "endpoint", reason: "failed to lookup address information: Temporary failure in name resolution" })
RUST_BACKTRACE=1
environment variable to display a backtrace Traceback (most recent call last): /opt/savant/src/module/run.py:5