NetHosted Community

Hi,

As many of your are aware there was a downtime on Neptune this morning. This took a while to be picked up on due to two main issues:

1) The tech supposed to be online at this time wasn't. This has been dealt with internally.

2) Both monitoring systems we have (our own, and a third party) didn't pick up on the downtime as explained below.

This downtime appears to have been caused by a firmware bug in the RAID card on this server. While under heavy I/O load the RAID card started to queue commands instead of actioning them leading to a situation where it basically gave up due to the number of queued commands. This caused the filesystems to be mounted as read only and this is what caused the two monitoring systems not to pick up on the downtime. They both check for the availability of services not what the services actually return. So although httpd was down as far as you were all concerned (as it was showing forbidden) as far as the monitoring software was concerned it was up as it was still responding.

We control one of these monitoring solutions and on this changes will be made to pick up on this specific edge case.

Tonight action will be taken to fix this bug with the RAID card. We may also re-flash the BIOS as there is one potential bug with our current version that may have also contributed to this. There will be a small amount of downtime while these processes are taking place.

Please accept our apologies for this unusual downtime and rest assured we are putting in place multiple measures to stop this from happening again (including a certain member of staff investing in multiple alarm clocks).

Thanks,

Andrew

_________________
| Andrew Bassett
| Managing Director, NetHosted Ltd.
| Earn £30 every time you refer someone to us!
| Follow us on Twitter:
| Members, tell us what of NetHosted!