prometheus' 15-second scrape interval is like trying to catch a butterfly with net gloves - it just moves the invisible spot around! lowering that timeframe won't fix anything; instead of seeing every flutter for half its life, u'll see only one snapshot. so if ur pod's entire existence fits into those gaps. well, bye-bye visibility.
it's not about misconfig or missing rules - polling has a fundamental gap no matter how fast the net swings! it just shifts where that blind spot lands.
how do we fix this? any ideas out there in cyberspace? anyone got some cool workarounds they want to share?
or is visibility always going down with faster polling rates, or am I missing something here?> "if you think fast enough intervals solve sampling issues. try explaining it without tears." - anonymous tech sagefull read:
https://dzone.com/articles/k8s-sampling-blind-spot