I'm using Puppeteer-cluster to process a high volume of screenshots directly from various HTML strings, and the response should be ASAP in a few milliseconds.
Since opening and closing a browser for each screenshot is not efficient, we want to use a cluster of 10 browsers, and we saw that it gave us the wanted results in the processing speed.
My concerns are from memory piling up and I want to ask if any known best practices can be used to take care of that?
E.g:
- Can the browsers in the cluster be refreshed?
- Should we close the page after each screenshot? etc...
Thanks!