chrome binary for lambda?

[andreabisello]

hi for running this project on lambda is required a particular version of the chrome binary? where have you found the binary contained in the lambda node release? thanks

[dejanvasic85]

here here! Can someone answer this one? PLease...

[wesmcouch]

Which binary do you need? We have chrome 62, and 68 are confirmed to be good versions. These are the full versions of chrome that work in Lambda.

If you want to download them, use the aws cli with one of these commands aws s3api get-object --bucket bb-test-resources --key chrome-68.0.3440.75.tar.gz --request-payer requester chrome-68.0.3440.75.tar.gz aws s3api get-object --bucket bb-test-resources --key chrome-62.0.3202.94.tar.gz --request-payer requester chrome-62.0.3202.94.tar.gz

These binaries are custom built and require quite a few settings to get running, this repo is a bit behind with what we have running currently.

[wesmcouch]

Here is a list of required arguments that need to be sent to the chromedriver --disable-gpu --single-process --no-sandbox --data-path=/tmp/data-path --homedir=/tmp/homedir --disk-cache-dir=/tmp/cache-dir --allow-file-access-from-files --disable-web-security --disable-extensions --ignore-certificate-errors --disable-ntp-most-likely-favicons-from-server --disable-ntp-popular-sites --disable-infobars --disable-dev-shm-usage

[wesmcouch]

At runtime, you should extract the tar to /tmp directory so that it is in the correct location. LD_LIBRARY_PATH environment variable should be /tmp/chrome-68.0.3440.75/libs:/lib64:/usr/lib64:/var/runtime:/var/runtime/lib:/var/task:/var/task/lib

[tsu-denim]

That binary can be found here as well... https://github.com/tsu-denim/chromium-full-gtk2-lambda https://github.com/tsu-denim/chromium-full-gtk2-lambda/releases

[andreabisello]

thanks

[tsu-denim]

I usually just rename the binary to just 'chrome' so I don't have to change the configs for chromedriver and LD_LIBRARY_PATH after upgrading... https://github.com/tsu-denim/strafer-duty/blob/master/pipeline/integration_tests/protractor.conf.js#L102

[andreabisello]

Here is a list of required arguments that need to be sent to the chromedriver --disable-gpu --single-process --no-sandbox --data-path=/tmp/data-path --homedir=/tmp/homedir --disk-cache-dir=/tmp/cache-dir --allow-file-access-from-files --disable-web-security --disable-extensions --ignore-certificate-errors --disable-ntp-most-likely-favicons-from-server --disable-ntp-popular-sites --disable-infobars --disable-dev-shm-usage

what about disk-cache-dir ? chrome on lambda has cache?

[wesmcouch]

Yeah, you must specify this or Chrome will try to put it in some directory that isn't writable in lambda. Before each test at the start of the lambda function you want to clear /tmp directory to refresh the environment

[tsu-denim]

Here is an example in Python...

https://github.com/tsu-denim/strafer-duty/blob/master/functions/lib/binary_service.py

Like Wes said, you should run both a function to wipe out /tmp and killing leftover processes before launching Chromedriver and Chrome. Chrome is pig for cache and will fill up tmp in no time, especially for modern client driven React/angular single page applications. Chromedriver tends to stick around when it is supposed to exit, so you want to check for stuff like that too. We found this through a lot of trial and error and it keeps things running in a predictable manner.