We are experiencing a network performance degradation for some servers at our uk-cov1 location. We are investigating the cause.
Following a process of elimination, we were able to identify a network switch as the root cause. The switch chip (ASIC) is intermittently dropping a selection of "jumbo frame" packets when certain configuration conditions are met. We were able to rework our configuration to avoid this issue at 05:50 UTC. We have notified the equipment vendor. This switch is part of a new rack deployment which affected only two host nodes carrying three virtual servers. Nonetheless, we apologise for any inconvenience caused.