Issues with Running a Large-Scale Google Sheets to GPT to Google Sheets Update Row Scenario

[Ecommike]

Hello everyone,

I’m currently working on a project where I’m utilizing the “Google Sheets Watch New Rows” to “GPT” to “Google Sheets Update a Row” scenario on the make.com platform. My Google Sheet contains several thousand rows, and I’m encountering a recurring issue: after processing about 100 rows (the number slightly varies), the scenario stops updating the rows. Although the scenario continues to run and doesn’t show any errors, no further rows are updated. To resolve the issue, I have to manually restart the scenario from the last row where it stopped.

This behavior is quite obstructive, especially when processing large amounts of data, and I wonder if there’s a way to configure the scenario or set up an error handling mechanism so it continuously processes all rows without manual intervention. Has anyone here experienced something similar or knows a solution to this problem?

Are there specific settings or methods in error handling or within the scenario settings that I could adjust to ensure the scenario runs smoothly without stopping after a certain number of rows? Any tips or advice would be greatly appreciated as I aim to avoid manual restarts and improve the efficiency of my workflow.

Thank you in advance for your support and advice!

[Nathan_Kind]

HI there @Ecommike.

You could try and use a repeater. I believe you should be able to choose which bundles of data to process using a filter and variable from the repeater (eg only bundles 1-100 where 1 is i so it would look like i-i00). At the end of your scenario have the variable go up by 1 and then when the repeater runs through the bundles again the filter will only let the next 100 pass through. Set the repeater to run enough time to get through your rows. It would look something like this:

With some formatting that may help solve the issue.

I’d be curious to know what the final output of wither chat GPT or Google Sheets was when it failed. Sometimes the final bundle of data displays what the error was is there was one.