As well as apparently multiple hundreds of …negatively…? executed CREATE, ALTER or DROP commands … averaging out to 0 per second.
… resulting in a spike of negative, I dunno, 30mb/s of reads.
Mhm yep.
My initial guess is that failed/malformed attempts to … do something … are being registered as negative instead of being actually displayed as failed.
But that seems to make too much sense.
Second guess: Lightning strike.
Third guess: Arcane, hitherto unseen bug or series of bugs utterly and totally unique to this exact set up which has somehow never occurred or been noticed before.
Is this saying network traffic was negative?
As well as apparently multiple hundreds of …negatively…? executed CREATE, ALTER or DROP commands … averaging out to 0 per second.
… resulting in a spike of negative, I dunno, 30mb/s of reads.
Mhm yep.
My initial guess is that failed/malformed attempts to … do something … are being registered as negative instead of being actually displayed as failed.
But that seems to make too much sense.
Second guess: Lightning strike.
Third guess: Arcane, hitherto unseen bug or series of bugs utterly and totally unique to this exact set up which has somehow never occurred or been noticed before.
Probably a traffic_bytes_counter got reset. You can see a lot of graphs went negative at the same time, so something probably restarted.
Metrics software like Prometheus will handle counter resets correctly for graphs like this.