error on Raspberry pi for tfjs-node

[lijianyu1985]

System information

• OS Platform and Distribution (e.g., Linux Ubuntu 16.04): ubuntu 22.04
• Mobile device : raspberry pi 4b
• TensorFlow.js installed from (npm or script link): npm i @tensorflow/tfjs-node
• TensorFlow.js version: 3.19.0

Describe the problem

1. throw error: cannot open shared object file: No such file or directory
2. and "npm rebuild @tensorflow/tfjs-node --build-from-source" stuck

Any other info / logs

[mattsoulanille]

Thanks for the bug report, @lijianyu1985. I'm not sure what's causing this, and I'm unable to reproduce it on my x86_64 machine. I'll try to reproduce it on a Raspberry pi, but in the meantime, I have some questions and things you can try.

1. What version of NodeJS are you using?
2. Can you try requiring the tfjs_binding.node file directly? To try this, move to the directory containing it, run Node, and then run require('./tfjs_binding.node').
3. What is the output of ldd tfjs_binding.node? If you don't have ldd, you can use readelf -d tfjs_binding.node instead.

[google-ml-butler[bot]]

Closing as stale. Please @mention us if this needs more attention.

[google-ml-butler[bot]]

Are you satisfied with the resolution of your issue? Yes No

[lijianyu1985]

Thanks for the bug report, @lijianyu1985. I'm not sure what's causing this, and I'm unable to reproduce it on my x86_64 machine. I'll try to reproduce it on a Raspberry pi, but in the meantime, I have some questions and things you can try.

1. What version of NodeJS are you using?
2. Can you try requiring the tfjs_binding.node file directly? To try this, move to the directory containing it, run Node, and then run require('./tfjs_binding.node').
3. What is the output of ldd tfjs_binding.node? If you don't have ldd, you can use readelf -d tfjs_binding.node instead.

Hi @mattsoulanille , Thanks for your response, I fixed this issue finally, it's a network issue.

[norbertnemeth]

Thanks for the bug report, @lijianyu1985. I'm not sure what's causing this, and I'm unable to reproduce it on my x86_64 machine. I'll try to reproduce it on a Raspberry pi, but in the meantime, I have some questions and things you can try.

1. What version of NodeJS are you using?
2. Can you try requiring the tfjs_binding.node file directly? To try this, move to the directory containing it, run Node, and then run require('./tfjs_binding.node').
3. What is the output of ldd tfjs_binding.node? If you don't have ldd, you can use readelf -d tfjs_binding.node instead.

Hi @mattsoulanille , Thanks for your response, I fixed this issue finally, it's a network issue.

@lijianyu1985 What was the issue? I'm stuck with the same error.

Thanks for the bug report, @lijianyu1985. I'm not sure what's causing this, and I'm unable to reproduce it on my x86_64 machine. I'll try to reproduce it on a Raspberry pi, but in the meantime, I have some questions and things you can try.

1. What version of NodeJS are you using?
2. Can you try requiring the tfjs_binding.node file directly? To try this, move to the directory containing it, run Node, and then run require('./tfjs_binding.node').
3. What is the output of ldd tfjs_binding.node? If you don't have ldd, you can use readelf -d tfjs_binding.node instead.

@mattsoulanille I hope these helps to figure out what's happening:

Context: Docker (node:16-buster-slim) OS: Debian GNU/Linux 10 (buster) Node version: v16.17.0 Package: @tensorflow/tfjs-node Installed packages with apt-get:

apt-get install -y build-essential \
    wget \
    python3 \
    make \
    gcc \ 
    libc6-dev \
    libcairo2-dev \
    libpango1.0-dev \
    libjpeg-dev \
    libgif-dev \
    librsvg2-dev \

node@4eae90eab5d7:/home/site/wwwroot/node_modules/@tensorflow/tfjs-node/lib/napi-v8$ ls -la
total 96
drwxr-xr-x 2 root root  4096 Aug 23 11:13 .
drwxr-xr-x 3 root root  4096 Aug 23 11:13 ..
-rwxr-xr-x 1 root root 87272 Jul 22 22:53 tfjs_binding.node
node@4eae90eab5d7:/home/site/wwwroot/node_modules/@tensorflow/tfjs-node/lib/napi-v8$ node
Welcome to Node.js v16.17.0.
Type ".help" for more information.
> require('./tfjs_binding.node')
Uncaught:
Error: /home/site/wwwroot/node_modules/@tensorflow/tfjs-node/lib/napi-v8/tfjs_binding.node: cannot open shared object file: No such file or directory
    at Object.Module._extensions..node (node:internal/modules/cjs/loader:1210:18)
    at Module.load (node:internal/modules/cjs/loader:1004:32)
    at Function.Module._load (node:internal/modules/cjs/loader:839:12)
    at Module.require (node:internal/modules/cjs/loader:1028:19)
    at require (node:internal/modules/cjs/helpers:102:18) {
  code: 'ERR_DLOPEN_FAILED'
}

node@4eae90eab5d7:/home/site/wwwroot/node_modules/@tensorflow/tfjs-node/lib/napi-v8$ ldd tfjs_binding.node
        not a dynamic executable
node@4eae90eab5d7:/home/site/wwwroot/node_modules/@tensorflow/tfjs-node/lib/napi-v8$ readelf -d tfjs_binding.node

Dynamic section at offset 0x10d80 contains 32 entries:
  Tag        Type                         Name/Value
 0x0000000000000001 (NEEDED)             Shared library: [libtensorflow.so.2]
 0x0000000000000001 (NEEDED)             Shared library: [libtensorflow_framework.so.2]
 0x0000000000000001 (NEEDED)             Shared library: [libstdc++.so.6]
 0x0000000000000001 (NEEDED)             Shared library: [libm.so.6]
 0x0000000000000001 (NEEDED)             Shared library: [libgcc_s.so.1]
 0x0000000000000001 (NEEDED)             Shared library: [libpthread.so.0]
 0x0000000000000001 (NEEDED)             Shared library: [libc.so.6]
 0x000000000000000e (SONAME)             Library soname: [tfjs_binding.node]
 0x000000000000001d (RUNPATH)            Library runpath: [$ORIGIN/../../deps/lib]
 0x000000000000000c (INIT)               0x4000
 0x000000000000000d (FINI)               0xe3cc
 0x0000000000000019 (INIT_ARRAY)         0x11d20
 0x000000000000001b (INIT_ARRAYSZ)       24 (bytes)
 0x000000000000001a (FINI_ARRAY)         0x11d38
 0x000000000000001c (FINI_ARRAYSZ)       8 (bytes)
 0x000000006ffffef5 (GNU_HASH)           0x260
 0x0000000000000005 (STRTAB)             0x13e0
 0x0000000000000006 (SYMTAB)             0x408
 0x000000000000000a (STRSZ)              6531 (bytes)
 0x000000000000000b (SYMENT)             24 (bytes)
 0x0000000000000003 (PLTGOT)             0x12000
 0x0000000000000002 (PLTRELSZ)           3576 (bytes)
 0x0000000000000014 (PLTREL)             RELA
 0x0000000000000017 (JMPREL)             0x3108
 0x0000000000000007 (RELA)               0x2f28
 0x0000000000000008 (RELASZ)             480 (bytes)
 0x0000000000000009 (RELAENT)            24 (bytes)
 0x000000006ffffffe (VERNEED)            0x2eb8
 0x000000006fffffff (VERNEEDNUM)         3
 0x000000006ffffff0 (VERSYM)             0x2d64
 0x000000006ffffff9 (RELACOUNT)          8
 0x0000000000000000 (NULL)               0x0

[norbertnemeth]

Repro steps:

1. Docker file

FROM public.ecr.aws/docker/library/node:16-buster-slim

RUN apt-get update && \ 
    apt-get install -y build-essential \
    wget \
    python3 \
    make \
    gcc \ 
    libc6-dev \
    libcairo2-dev \
    libpango1.0-dev \
    libjpeg-dev \
    libgif-dev \
    librsvg2-dev

WORKDIR /home/site/wwwroot
RUN npm install @tensorflow/tfjs-node
CMD [ "echo hello" ]

2. Start the container, go to the specific folder, run node and try to import the tfjs_binding.node file:

root@3ac3d6a6a3c3:/home/site/wwwroot# cd /home/site/wwwroot/node_modules/@tensorflow/tfjs-node/lib/napi-v8
root@3ac3d6a6a3c3:/home/site/wwwroot/node_modules/@tensorflow/tfjs-node/lib/napi-v8# node
Welcome to Node.js v16.17.0.
Type ".help" for more information.
> require('./tfjs_binding.node')
Uncaught:
Error: /home/site/wwwroot/node_modules/@tensorflow/tfjs-node/lib/napi-v8/tfjs_binding.node: cannot open shared object file: No such file or directory
    at Object.Module._extensions..node (node:internal/modules/cjs/loader:1210:18)
    at Module.load (node:internal/modules/cjs/loader:1004:32)
    at Function.Module._load (node:internal/modules/cjs/loader:839:12)
    at Module.require (node:internal/modules/cjs/loader:1028:19)
    at require (node:internal/modules/cjs/helpers:102:18) {
  code: 'ERR_DLOPEN_FAILED'
}

[lijianyu1985]

Hi @norbertnemeth , In my case the issue is that the tfjs_binding.node file is downloaded from cloud, but the downloaded file is not fit for my platform (raspberry pi), So I rebuild it for my platform then it worked. So, I think you can rebuild that file in your docker

how to rebuild: https://github.com/tensorflow/tfjs/tree/master/tfjs-node#rebuild-the-package-on-raspberry-pi

[norbertnemeth]

Thanks @lijianyu1985 , it resolved my issue too.

[WillAvudim]

Ran into the same issue and I confirm that

npm rebuild @tensorflow/tfjs-node --build-from-source

fully resolves it and lets me run on AWS ARM CPUs.

[TobiasFriberg]

Ran into the same issue and I confirm that

npm rebuild @tensorflow/tfjs-node --build-from-source

fully resolves it and lets me run on AWS ARM CPUs.

THIS^

thank you so much, worked like a charm!