Edge functions performance degradation

[cers]

Across our app, all of our APIs have degraded at the same time, several hours after any deploys had taken place. In case it’s relevant, we’re using next@14.2.13

Here is an example from some of them:

As can be clearly seen, around mid-day on the 16th of October, there was a sharp increase in CPU time.
Our functions use the Basic CPU, but that setting has not changed for at least months.

Any ideas what could be causing this?

Deployment URL or Custom Domain:
Environment (local, preview, production):
Project Framework: nextjs
Build Settings:
  Framework Preset:
  Build Command (if not default): next build --no-lint
  Output Directory (if not default):
  Install Command (if not default):
Node/Runtime Version: node v20
Package Manager: npm
Relevant Packages:

[amyegan]

Thanks for reporting this, @cers. Looking into it now

[amyegan]

I want to keep you updated on what I found so far.

There were a couple of incidents reported around the time you observed an increase in CPU time. They affected Edge Functions but have since been resolved.

• Increased INTERNAL_EDGE_FUNCTION_INVOCATION_FAILED Errors
• Elevated 5xx errors in cpt1 and fra1 regions

It looks like ISR reads increased around the same time. I don’t mean to imply that the two are related, but there may have been a change around that time which affected both metrics. Something as small as a memory configuration change or a new npm package could have a noticeable effect.

Please let me know if you think of anything that changed at, or just before, that time.

[cers]

The last deploys we had before that were around 4AM, and were minor updates to a react library. I cannot imagine that is related?

Also, the slowdown has persisted since, so it doesn’t appear to be a transient issue