Things kept breaking, and it wasn’t because I wrote bad code.
It was IAM.
I was setting up a cloud lab in OCI and everything looked fine on the surface, instances, networking, policies.
But every time I tried to launch or manage resources, something failed.
Not loudly. Just enough to slow everything down.
At first, I assumed it was a config issue.
So I kept switching between tasks, checking networking, then compute, then scripts.
That made it worse.
I wasn’t losing time… I was losing context.
What finally worked was slowing down and tracing the failure properly.
I started reading the exact error messages and mapping them back to IAM policies.
That’s when I finally saw it.
A small missing permission:
Allow group to read instance-images in tenancy
Once I added it, everything started working instantly.
No changes to code. No changes to architecture.
Just IAM.
Nothing else changed.
The biggest lesson?
Most cloud failures aren’t about what you built, they’re about what you forgot to allow.
If something feels randomly broken in the cloud, start with IAM first.
Top comments (0)