RunPod•6mo ago

JS pupeteer issue

Hi, I have a pupeteer chromium task that i want to run on the runpod which is written in typescript. I did not see any JS handler. So, i am calling the JS entrypoint via python as below.

""" Example handler file. """

import runpod
from subprocess import call
import os

# If your handler runs inference on a model, load the model here.
# You will want models to be loaded into memory before starting serverless.


def handler(job):
    """ Handler function that will be used to process jobs. """
    job_input = job['input']

    prompt_id = job_input.get('aiPromptId')
    video_save_path = job_input.get('videoSavePath')
    
    # Create a copy of the current environment variables
    env = os.environ.copy()

    # Add or update the required environment variables
    env['AI_PROMPT_ID'] = prompt_id
    env['VIDEO_SAVE_PATH'] = video_save_path
    print("env", env)
    
    print("Running the executeDockerCommand")
    # env['NODE_ENV'] = 'development'

    call(["yarn", "ts-node", "src/backend/executeDockerCommand.ts"], env=env)

    print("Finish running the command")

    return f"Hello, {prompt_id}!"


runpod.serverless.start({"handler": handler})

""" Example handler file. """

import runpod
from subprocess import call
import os

# If your handler runs inference on a model, load the model here.
# You will want models to be loaded into memory before starting serverless.


def handler(job):
    """ Handler function that will be used to process jobs. """
    job_input = job['input']

    prompt_id = job_input.get('aiPromptId')
    video_save_path = job_input.get('videoSavePath')
    
    # Create a copy of the current environment variables
    env = os.environ.copy()

    # Add or update the required environment variables
    env['AI_PROMPT_ID'] = prompt_id
    env['VIDEO_SAVE_PATH'] = video_save_path
    print("env", env)
    
    print("Running the executeDockerCommand")
    # env['NODE_ENV'] = 'development'

    call(["yarn", "ts-node", "src/backend/executeDockerCommand.ts"], env=env)

    print("Finish running the command")

    return f"Hello, {prompt_id}!"


runpod.serverless.start({"handler": handler})

The above code works fine locally. But, when i deploy to serverless, it kicks off the yarn ts-node ... , but gets stuck and does not emit any logs and eventually times out. Any advice?

25 Replies

nerdylive•6mo ago

Use codes block it's better to write code inside a code block This is how you write block code: ``` import subprocess import runpod ``` It will be like:

import subprocess
import runpod

import subprocess
import runpod

It does emit logs, if it does in your local when you call it manually in your command line But the logs from your process(cli you call using subprocess didn't go through runpod's log) So what you have to do is:

from runpod import RunPodLogger
# change this to your command ( also put this inside a handler code )
x = subprocess.run(["cmd", "/c", "dir"], capture_output=True)
# initialize the logger object
logger = RunPodLogger()
# log the output from your process
logger.info(f'{x.stdout.decode()}')

from runpod import RunPodLogger
# change this to your command ( also put this inside a handler code )
x = subprocess.run(["cmd", "/c", "dir"], capture_output=True)
# initialize the logger object
logger = RunPodLogger()
# log the output from your process
logger.info(f'{x.stdout.decode()}')

ACOP•6mo ago

Awesome! Let me try that in the morning.

ACOP•6mo ago

Looks like still stuck and does not emit log or proceeds

nerdylive•6mo ago

It does It says it couldn't read that folder

ACOP•6mo ago

i take that back. Its pupetteer launch issue. I am launching pupeteer with these options

        const puppeteerLaunchOpts: PuppeteerLaunchOptions = {
                // headless: shouldPerformanceTrace() ? false : 'shell',
                headless: true,
                devtools: isDev,
                 args: ['--disable-gpu', '--no-sandbox', '--disable-setuid-sandbox'],
        };

    const browser = await puppeteer.launch(puppeteerLaunchOpts)

        const puppeteerLaunchOpts: PuppeteerLaunchOptions = {
                // headless: shouldPerformanceTrace() ? false : 'shell',
                headless: true,
                devtools: isDev,
                 args: ['--disable-gpu', '--no-sandbox', '--disable-setuid-sandbox'],
        };

    const browser = await puppeteer.launch(puppeteerLaunchOpts)

Looks like this gets stuck

nerdylive•6mo ago

I c

ACOP•6mo ago

It works fine locally 🤔 i am ssh'ing into the serverless and still seeing the same issue

nerdylive•6mo ago

whats the error? can you copy in text? maybe i can help

ACOP•6mo ago

its just stuck

nerdylive•6mo ago

stuck in what? any logs?

ACOP•6mo ago

There are no logs being emitted by puppeteer.launch I added logs before and after puppeteer.launch and the log after launch doesnt get triggered.

ACOP•6mo ago

nerdylive•6mo ago

is that warn normal or what

ACOP•6mo ago

yea. thats normal it emits locally as well. Unrelated Has anyone ran pupeteer successfully?

nerdylive•6mo ago

hmm i dont know about that send me the full file maybe let me see

ACOP•6mo ago

sure. Let me isolate keys and just keep the pupeteer code

ACOP•6mo ago

This is the entire file

message.txt

ACOP•6mo ago

The entry point is generateAndUploadVideoViaPuppeteer

nerdylive•6mo ago

try dumpio = true on launchBrowser put it in args

ACOP•6mo ago

got it! trying

nerdylive•6mo ago

whats the logs after

ACOP•6mo ago

Got this now Is there something i need to turn on gpu support via dockerfile?

nerdylive•6mo ago

no i dont think so search the problem up i think its like dependency problem only

ACOP•6mo ago

Got it. Let me search around.

Gaming

Programming

JS pupeteer issue