|
upstream contribution · merged ✓
silent failure in the unidling controller
github.com/ovn-kubernetes/ovn-kubernetes · PR #6558
|
|
🐣 first — what is "unidling"?
OpenShift can scale services to zero when unused.
When traffic arrives → the service "unidles" (wakes up).
The unidling controller handles this wake-up process.
It watches for traffic events and triggers pod scale-up.
|
|
|
🐛 three problems found
|
1. typo in log message
"ContollerEvent"
→ should be
"ControllerEvent"
makes log searching unreliable — grep finds nothing
|
|
2. typo in comment
"reconcilation"
→ should be
"reconciliation"
|
|
3. 🔴 silent nil error — the real bug
when the "vip" key is missing from an event,
the function returned nil instead of an error.
failure silently swallowed → unidling silently broken
|
|
|
|
💭 why does returning nil matter?
|
before fix
event arrives
"vip" key missing
↓
return nil ← "all good!"
↓
caller: no retry
↓
service stays idle 😶
|
|
after fix
event arrives
"vip" key missing
↓
return error ← "something's wrong"
↓
caller: retries + logs
↓
visible, debuggable ✓
|
|
|
|
💡 the rule
never return nil when something went wrong.
Silent failures are the hardest bugs to debug —
the system looks healthy while quietly doing nothing.
If something unexpected happens → say so.
An error that surfaces is always better than one that hides.
|
|
|
merged: Jun 17 2026 |
repo: ovn-kubernetes/ovn-kubernetes |
pr: #6558
small fix · big lesson · contributed by Parikshit Khedekar
|
|