Have a dedicated channel (or template) to report suspected call-caching issues

Post author
Steve Huang
  • Edited

From time to time, I experience issues with call-caching.

The most annoying is when a heavy workflow fails due to some transient errors in trivial—but later—stages, with the most expensive parts already finished successfully. Yet when relaunched with call-caching enabled, the expensive parts are re-run for unknown reasons: we are not sure if it was a cache-write fail in the 1st run, or cache-read miss in the rerun. The end result is longer wait time and unnecessary cost.

But it takes time for users to report such observations. I feel many times because of this extra burden on an already frustrated user, we simply don't report them and eat the costs. 

This can be remediated somehow by providing a template or dedicated channel for reporting these apparent call-caching issues. Obviously not everything is going to be legit call-caching bugs. However, the accumulated data will help catch bugs, if any. 

Thank you.

Steve

Comments

0 comments

Please sign in to leave a comment.