ChromiumProcessException: Failed to launch Chromium!

[DevGenesis95]

Description

When running the function in AWS lambda getting this exception.

{ "ClassName": "PuppeteerSharp.ChromiumProcessException", "Message": "Failed to launch Chromium! ", "Data": null, "InnerException": null, "HelpURL": null, "StackTraceString": " at PuppeteerSharp.ChromiumProcess.State.StartingState.StartCoreAsync(ChromiumProcess p)\n at PuppeteerSharp.ChromiumProcess.State.StartingState.StartCoreAsync(ChromiumProcess p)\n at PuppeteerSharp.Launcher.LaunchAsync(LaunchOptions options)\n at PuppeteerSharp.Launcher.LaunchAsync(LaunchOptions options)\n at ConvertorHelpers.PdfConvertor.ConvertToPdf2(String url, String fileName, String filePath) in ConvertorHelpers\PdfConvertor.cs:line 107", "RemoteStackTraceString": null, "RemoteStackIndex": 0, "ExceptionMethod": null, "HResult": -2146233088, "Source": "System.Private.CoreLib", "WatsonBuckets": null }

Complete minimal example reproducing the issue

The executable path which is formed in lambda function after fetching the browser is "/tmp/chromium/Linux-706915/chrome-linux/chrome"

E.g.

// Arrange
 var options = new LaunchOptions
                {
                    ExecutablePath = executablePath,     
                    Headless = true,
                    Devtools = false,
                    LogProcess = true,
                    Args = new[] { "--no-sandbox" }
                };

var chromiumRevision = BrowserFetcher.DefaultRevision;
  using (var browser = await Puppeteer.LaunchAsync(options)) // Exception line 107

// Act
...

// Assert
...

Expected behavior:

launch and convert HTML to PDF

Actual behavior:

Crasehs

Versions

• Which version of PuppeteerSharp are you using? -- 2.0.3
• Which .NET runtime and version are you targeting? E.g. .NET framework 4.6.1 or .NET Core 2.0. -- .Net core 2.1 in AWS lambda

Additional Information

Its working fine when m executing it locally i.e. on my PC(windows) but on lambda function when running on Linux getting the error.

[kblok]

Hey @DevGenesis95 can you take a look at this https://github.com/litmus/HeadlessChromium.Puppeteer.Lambda.Dotnet and see if that helps?

[DevGenesis95]

Thanks for the reply @kblok but I have tried that too but unfortunately it is not compiling with my lambda which is in .net core 2.1 as mentioned in the repo. Apart from that, I have tried to execute all the permission through my code but yet the issue still persists.

            var fileInfoexec = new UnixFileInfo(executablePath);
            fileInfoexec.FileAccessPermissions = FileAccessPermissions.AllPermissions |
                                             FileAccessPermissions.AllPermissions;

[kblok]

@DevGenesis95 I think it's not only about a permission fix but which is the chromium-browser AWS supports, see https://github.com/alixaxel/chrome-aws-lambda/tree/master/bin

[DevGenesis95]

Umm Actually I don't know much about node js..are you taking about the browser revisions? Is that the issue.

[kblok]

It seems AWS won't be able to run a normal chromium release. Those 2 libraries are downloading another chromium version there.

[DevGenesis95]

The nuget used is PuppeteerSharp version 2.0.3 and the Browserfetcher.DefaultRevision = 706915. Should I degrade the puppeteer version? Any suggestion about what should I do. It would be really appreciated :)

[kblok]

I created this issue https://github.com/litmus/HeadlessChromium.Puppeteer.Lambda.Dotnet/issues/49 :)

[DevGenesis95]

Thanks for the help 👍

[brianfeucht]

@DevGenesis95 you can specify the version of PuppeterSharp you want to use in your csproj file. You do this by adding a direct reference to the Nuget package instead of inheriting the minimum version set in HeadlessChromium.Puppeteer.Lambda.Dotnet

[DevGenesis95]

@brianfeucht Yes, I have downgraded the Pupeeter NuGet package to 1.14.1 which uses the DefaultRevision = 641577 but still, the error persists. Attaching cloudwatch screenshot for this

[brianfeucht]

@DevGenesis95 HeadlessChromium.Puppeteer.Lambda.Dotnet build 1.0.1.39 is chromium 756035 (83.0.4103.0)

I am running this with PuppeteerSharp build 2.0.3 on Lambda .NET 3.1. It should work with Lambda .NET 2.1 but I haven't verified that myself. Would you be willing to try that combo?

[DevGenesis95]

Thanks for the info actually I can go with .Net 3.1 if that's working. No, it's not working in 2.1 for me. Can you give me a solution where you have integrated it 3.1 lambda that would be really helpful :)

[brianfeucht]

Our project files looks like this:

<?xml version="1.0" encoding="utf-8"?>
<Project Sdk="Microsoft.NET.Sdk">
  <PropertyGroup>
    <TargetFramework>netcoreapp3.1</TargetFramework>
    <GenerateRuntimeConfigurationFiles>true</GenerateRuntimeConfigurationFiles>
  </PropertyGroup>
  <ItemGroup>
    <None Remove="aws-lambda-tools-defaults.json" />
    <Content Include="aws-lambda-tools-defaults.json">
      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
    </Content>
  </ItemGroup>
  <ItemGroup>
    <PackageReference Include="PuppeteerSharp" Version="2.0.3" />
    <PackageReference Include="Amazon.Lambda.Serialization.Json" Version="1.2.0" />
    <PackageReference Include="HeadlessChromium.Puppeteer.Lambda.Dotnet" Version="1.0.1.39" />
  </ItemGroup>
</Project>

The aws-lambda-tools-default.json file looks like:

{
  "configuration" : "Release",
  "framework" : "netcoreapp3.1",
  "function-runtime":"dotnetcore3.1",
  "function-memory-size" : 1536,
  "function-timeout" : 60,
  "function-handler" : "Lambda::LambdaHandler::Handle"
}

We deploy following a process similar to this: https://aws.amazon.com/premiumsupport/knowledge-center/build-lambda-deployment-dotnet/

[DevGenesis95]

Ok but can you check mine..

I have created a lambda function(3.1) and referenced your puppeteer library in it and then run the PowerShell script.

The handler looks like this currently I'm resolving the dependencies and will upgrade the ram as well when i test it locally before publishing...for publishing, i use visual studio publish to lambda directly

[brianfeucht]

Ah I see what is going on here, there is a bit of magic in the creation of the nuget packages that make sure aws.tar.br, chromium.br and swiftshader.tar.br make into the final zip package uploaded to lambda. Since you are including this project this way, that magic is lost and those files do not make it into your Lambda deploy package.

Instead of using the project reference can you use the package reference?

[DevGenesis95]

Yes, I can use in any way been lost from few days :( ok joke apart

so u want me to include these in my csproj of lambda ryt?

 <PackageReference Include="PuppeteerSharp" Version="2.0.3" />
    <PackageReference Include="Amazon.Lambda.Serialization.Json" Version="1.2.0" />
    <PackageReference Include="HeadlessChromium.Puppeteer.Lambda.Dotnet" Version="1.0.1.39" />

But just a bit confused why we including the "HeadlessChromium.Puppeteer.Lambda.Dotnet" this package if pupeetersharp doing all the stuff.

[brianfeucht]

All HeadlessChromium.Puppeteer.Lambda.Dotnet does is repackage a build of chromium for Lambda and handles copying those files to a place executable by .NET. It is meant to be a simple helper library to make PuppeteerSharp easier to use. Building chromium for lambda is a bunch of work so we actually reuse the work done for a NodeJs project

[DevGenesis95]

ok that's really impressive work u did there.

Update I have created a new lambda(3.1)..can you verify if its correct this time.

[brianfeucht]

@DevGenesis95 yes this looks like it should work

[DevGenesis95]

oops..got this when on run i think its gonna take a while :(

[brianfeucht]

@DevGenesis95 not sure the mock lambda will work on windows. Since the chromium build is specific to a Linux Lambda AMI it is likely even if you get past this error you will get a bunch of failures trying to start chrome

[DevGenesis95]

ok, last shot I'm trying is to run this in the AWS environment.

[DevGenesis95]

Kudos guys..finally its working ..able to generate pdf from URL

Thanks a ton!! @brianfeucht @kblok :)

[kblok]

Awesome!! Thank you @brianfeucht!