Rackspace forced to reboot cloud over Xen bug

Rackspace is following Amazon Web Services to urgently patch and reboot its servers in all regions to deal with the mysterious Xen virtualisation layer bug.

All Standard, Performance 1 and 2 servers within the company's NextGen infrastructure will have to be restarted, Rackspace said in an email sent out over the weekend.

OnMetal servers - which don't run Xen - will not need to be rebooted.

The reboot "at the hypervisor level" affects all servers, both Windows and Linux, Rackspace said.

In Australia, Rackspace will reboot its Sydney cloud servers on Tuesday September 30, during a three-hour window scheduled to start at 2100 hours AEST and end at one minute past midnight the following day.

On its system status page, Rackspace advised customers to ensure their services are configured to start up again after the servers are booted, and also to ensure that there are up-to-date server images and file-level backups of critical data.

Rackspace also asked customers to save any configuration changes ahead of the reboots.

Amazon Web Services said it would patch and restart its Elastic Compute Cloud (EC2) three days ago to deal with the Xen XSA-108 bug.

The nature of the XSA-108 bug is still under embargo until Thursday Australian time, which AWS said allows it to follow security best practices.

Rackspace said on its community forum that it is not currently aware of any instances in which the Xen issue has impacted customer data.

Update: Rackspace has clarified that the Australian reboots will in fact start on Tuesday September 30 at 2100 hours AEST, finishing at one minute past midnight the following day, and not on October 1.

A company spokesperson apologised for any confusion.