How to change from port 50051 in Riva quick start

AI & Data Science Deep Learning (Training & Inference) Riva
1

Please provide the following information when requesting support.

Hardware - Jetson AGX Xavier
Operating System - Jetpack 5.0.1
Riva Version - 2.3.0

I would like to link rivaTTS with Audio2Face.
I followed this topic and tried to change the port.
I did riva_clean.sh and then changed config.sh to 50050, but the server will not start.
(I have confirmed that QuickStart works on port 50051.)

How can I change the port to something other than 50051 so that it will work with A2F?

2

Hi @t.seki.tsc

Thanks for your interest in Riva

Can you please share the log output of

bash riva_init.sh | tee riva_init_logs.txt

and

bash riva_start.sh | tee riva_start_logs.txt

Share with us the files

Thanks

3

Hello rvinobha

I share flies

config.sh (14.1 KB)
riva_init_logs.txt (11.0 KB)
riva_start_logs.txt (986 Bytes)

4

Hi @t.seki.tsc

Thanks for sharing the files

Apologies, need one more file,

While running
bash riva_start.sh
can you parallely open a new terminal, in that new terminal pleae run the below command
docker logs riva-speech | tee docker_riva_speech_logs.txt
and share the file with us

riva_start.sh and docker logs riva-speech to be run in parallel in two terminals

Thanks

5

I tried the command you taught me, but there was no content in docker_riva_speech.txt, so I’ll share what I saw in the terminal

I0909 09:55:54.286854     7 riva_server.cc:118] Using Insecure Server Credentials
E0909 09:55:54.290094     7 model_registry.cc:238] error: unable to get server status: failed to connect to all addresses
E0909 09:56:04.290828     7 model_registry.cc:238] error: unable to get server status: failed to connect to all addresses
I0909 09:56:14.303933     7 model_registry.cc:112] Successfully registered: citrinet-1024-en-US-asr-streaming for ASR
I0909 09:56:14.412210     7 model_registry.cc:112] Successfully registered: riva-punctuation-en-US for NLP
I0909 09:56:14.421353     7 model_registry.cc:112] Successfully registered: riva_intent_misty for NLP
I0909 09:56:14.564692     7 model_registry.cc:112] Successfully registered: riva-punctuation-en-US for NLP
I0909 09:56:14.573410     7 model_registry.cc:112] Successfully registered: riva_intent_misty for NLP
I0909 09:56:14.595467     7 model_registry.cc:112] Successfully registered: fastpitch_hifigan_ensemble-English-US-Female-1 for TTS
I0909 09:56:14.620448     7 riva_server.cc:158] Riva Conversational AI Server listening on 0.0.0.0:50051
W0909 09:56:14.620584     7 stats_reporter.cc:41] No API key provided. Stats reporting disabled.
6

Hi @t.seki.tsc

Apologies for the delay

I have updates from the team

Regarding the error
error: unable to get server status: failed to connect to all addresses

it is not port related we believe, one of the thing we doubt is there may be another instance of Riva running in the same machine or host

Please run
bash riva_clean.sh

and retry installation

After the above command, if if the issue persists, please make a backup/safe copy of docker container/volumes/images, Please remove all the containers

docker rm -f $(docker ps -a -q)

and retry installation

Please let us know if the above works

Thanks

7

@rvinobha

Thanks for the reply.

I have tried two methods but no improvement.
I tried re-doing the DL from ngc and it didn’t change.
The docker log is the same as in my previous post.

8

Thanks @t.seki.tsc

Apologies that it didn’t work out,

But we are pretty sure mostly that the issue is not with port change

I will check with the team again with your feedback,

One more ask,

can we also try to remove all the volumes

docker volume rm $(docker volume ls -q)

and images

docker rmi $(docker images -a -q)

and try one more time fresh install

NOTE: please make sure to take backup/precaution as it may remove all images/volumes present in the docker

Thanks

9

Thanks @rvinobha

docker volume ls -q didn’t work.

I ran it in the following order

docker rm -f $(docker ps -a -q)
docker rmi $(docker images -a -q)

I then tried reinstalling but could not get it to run on port 50050.

Here are the docker logs

I1007 05:00:59.410478 7 riva_server.cc:118] Using Insecure Server Credentials
E1007 05:00:59.414355 7 model_registry.cc:238] error: unable to get server status: failed to connect to all addresses
E1007 05:01:09.414934 7 model_registry.cc:238] error: unable to get server status: failed to connect to all addresses
I1007 05:01:19.430328 7 model_registry.cc:112] Successfully registered: citrinet-1024-en-US-asr-streaming for ASR
I1007 05:01:19.583987 7 model_registry.cc:112] Successfully registered: riva-punctuation-en-US for NLP
I1007 05:01:19.596801 7 model_registry.cc:112] Successfully registered: riva_intent_misty for NLP
I1007 05:01:19.777410 7 model_registry.cc:112] Successfully registered: riva-punctuation-en-US for NLP
I1007 05:01:19.788636 7 model_registry.cc:112] Successfully registered: riva_intent_misty for NLP
I1007 05:01:19.820200 7 model_registry.cc:112] Successfully registered: fastpitch_hifigan_ensemble-English-US-Female-1 for TTS
I1007 05:01:19.822788 7 model_registry.cc:112] Successfully registered: fastpitch_hifigan_ensemble-English-US-Male-1 for TTS
I1007 05:01:19.860913 7 riva_server.cc:158] Riva Conversational AI Server listening on 0.0.0.0:50051
W1007 05:01:19.860963 7 stats_reporter.cc:41] No API key provided. Stats reporting disabled.

10

Hi @t.seki.tsc

Apologies for the long wait,

Finally i have updates on this issue,

There was indeed a problem with different riva server port on embedded devices, this is fixed and will be available in future release

The fix is top update the riva_start.sh with the below highlighted change (highlighted in red with box)

image
image1760×220 9.88 KB

Code snippet for the highlighted

echo "Starting Riva Speech Services. This may take several minutes depending on the number of models deployed."
docker rm $riva_daemon_speech &> /dev/null
if [[ $riva_target_arch == "arm64" ]]; then
    docker_run_args="-p 8000:8000 -p 8001:8001 -p 8002:8002 -p 8888:8888 --device /dev/bus/usb --device /dev/snd $image_speech_api riva_server --riva-uri=0.0.0.0:$riva_speech_api_port $ssl_args"
else
    docker_run_args="-p 8000 -p 8001 -p 8002 $image_speech_api start-riva --riva-uri=0.0.0.0:$riva_speech_api_port --asr_service=$service_enabled_asr --tts_service=$service_enabled_tts --nlp_service=$service_enabled_nlp $ssl_args &> /dev/null"
fi

Thanks

11

Thanks for the reply @rvinobha

I ran riva_clean.sh and riva_init.sh.
And I ran riva_start.sh after rewriting it.

But the riva server did not work.
It stops here.

I1206 04:53:54.568471 142 grpc_server.cc:4421] Started GRPCInferenceService at 0.0.0.0:8001
I1206 04:53:54.569345 142 http_server.cc:3113] Started HTTPService at 0.0.0.0:8000
I1206 04:53:54.613123 142 http_server.cc:178] Started Metrics Service at 0.0.0.0:8002

Do you know of a version of riva that works with the rewrite of riva_start.sh that you taught me?

12

Hi @t.seki.tsc

Can you share you riva_start.sh file ?

Thanks

13

@rvinobha

I’ll send you the file.

Thanks
riva_start.sh (4.8 KB)

14

Hi @t.seki.tsc

Thanks for sharing the file, the changes you have made is perfect,
Can you please share the complete log of riva_start.sh which fails, with check with the internal team

Thanks