what happens when a workflow step fails? error handling question

error handling works way better than zapier/make. heres the deal: 1. retries - you can configure automatic retries on any node. set how many times to retry and the delay between attempts. great for flaky APIs 2. error branch - every node has an "on error" output you can connect to a separate branch. so you can catch the failure and route it to a slack notification, add a row to an error log table, send yourself an email, whatever 3. execution log - theres a full log of every workflow run. you can filter by status (success/failed/running) and click into any failed execution to see exactly which node failed and the full error response including headers and body this was a huge selling point for us too. the visibility into failures is night and day compared to zapier

this is exactly what i was hoping to hear. the error branch concept is perfect. going to set up error > slack notification on all our critical workflows

the execution log is a lifesaver. we had a workflow fail at 2am last week and i could see the full API response the next morning without having to reproduce the issue. in make i would have had zero context

+1 the error handling here is solid. one tip - set up a dedicated error notification workflow that all your other workflows route to on failure. saves you from configuring slack/email notifications in every single workflow

thank god for actual error messages. the number of times zapier just said "an error occurred" with no details was infuriating

something i learned the hard way - if you have retries enabled, make sure your workflow is idempotent. meaning if the same data goes through twice it shouldnt create duplicates. use upserts instead of inserts when writing to tables and check for existing records before creating them