Y̵̢̡̧̡̳̦͍̲͔͚̳̩͔̔̊̌̔̂́̐͟e̵̡̖̰͚̪͓̭̮͈̳̳̓̈́͆̄̀̆͂̕ṥ̷̺͔͓͇̆̿̀͒̄̇͛̋,̴͓͉̙͈͌̌̅̑̄̎͝͝ ̵̝͔̣̗̼͒̉̿͂͜a̸̢̡̲̼̜͍̭̙͉̠̜̝̰̗͒̈͐ͅ ̵̡̱̻̞͇̠̦͓͖̖̗̱̣̺͈͑̑̈́͂̚͟͠͝p̶̧̨̧̬̘̪̦̠͙̼̄̊̄̃͌̋͟͜͡r̸̦̤͉̥̯̝͍̬̫̠̄̀̿ͅô̸̧͙̜̘͚̜̱̦̿͗̍̊͒̍̍ṕ̷͎̮͇̝͎̈́̋͒̌̓̔̔̂͊̅̌͂̾̈́̚ë̶̠̪͉̼̬̺̭͇́̋̉̎͌̉͡͡͝ͅȓ̸̻̻͂̎́ ̶̨̫̺͖̯̫̐̂̋̆̈́̆͆̽̆͗͟͟͟͝͝͠d̷̢̗̙͙͕̲̺̫̺͇̤̾͌̈́͜͠͝ă̸̼̣̞̮̖̩̘̙̳̗̟͔̰͕͛́ͅṡ̵̨̢̢͚̖͕̭͖̹̥̱͖̮̘̗͈̽͂̐͌̈̅̓̈́̕ĥ̷̡̠͓̹̲̱̣͎̤̖̺̣͚̘͓̐̄̍̐͐͗̈́͋͆̓̆̕͜͝b̵̢͉̗̖͙̖̞̉͌̉̄͊͘ó̶͙̳̫̻̓̈͂̌͆̃̕͟ä̶̡̢̜̟͚͇̝͕̻̻̭́̃͝͝ṟ̵̥̝͍̐̈́̐̉͊̽̾̍̕̕͠d̸̲̫̊̓̽̐̽.̵̨̛̭͔̯͔̖̥̟͓̩̰͖̋͑̓͂͆͆̈́͌͜͝ ̴͓̟͋͜
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.
ż̸̬̪̾̔̏͝a̴̛̙͈̦̟͎̓͐͗͘͜l̴̙͈͙͍̓͒̆g̸̠͑͝ô̷̬̭͑̋͘ ̴̘̰͎̤̒̈́̅͐͜ǫ̵̻̭̯͎̒̍͘͠b̶̹̟͍͓̻̐͐́ŝ̸̭̦̰͘e̶̠̝͙̳̽̃̇̑͆r̴̜͇̼̯͈̈́͛̅̾͝v̶͉̲̥̈́̈́̓̽ę̸̡̧͇̊̓͂̑s̷̠̑̉̈́̂̂͠
