Who is affected

All the machines, virtual servers or not, and services hosted in the main cabinet. This includes hosting and email services.

When

August 26th 2008, between 19h00 and 21h00 EDT (-0400). Outages described below will occur between 20h00 and 21h00, EDT. However, we hope to limit those outages to 30 minutes (so between 20h00 and 20h30).

What will happen

The main switch will be replaced. This will provoque short network outages for each of the hosted servers.

Why

The current switch is full and show signs of weaknesses. We prefer to replace it before a complete outage.

How

See the rapport d'intervention (fr). Note that this outage will begin with a general outage affecting all servers when the core router will be replugged. Then every machine will be replugged one by one, which should provoke a few minutes of outage for each machine. It is also possible that this procedure fails to function properly and that we go forward with a quick and dirty unplugging and replugging of everything.

I object!

If this intervention is too problematic for you or your organisation, please let us know within 24h to see if we can arrange otherwise.

Also not that the sysadmin blog will be updated if the intervention is changed in any way or if we experience problems or delays.