Amazon Web Services rolled out an urgent patch to hosts causing a widespread maintenance reboot of EC2 compute instances over the next several days, ending September 30. The company had a similar event in December 2011, but a more substantial number of instances will be rebooted this time around.
AWS has notified customers by email, but the company has so far been tight-lipped about the reasons for the reboot. The reboot is occurring across all availability zones.
There is speculation that the company is patching for vulnerability, specifically a fix for open source hypervisor Xen. The reason will be revealed on October 1 after all hosts have been patched. If customers relaunch an instance before the maintenance, they will not be guaranteed a patched host.
Instance reboots to upgrade and apply patches are not uncommon, however the scale and short notice behind this case is causing some concern. The short notice is most likely because of security reasons, which would make the move imperative.
Popular cloud management platform RightScale’s CTO Thorsten Von Eicken wrote that whenever the company’s operations team receives a maintenance notice, they relaunch as soon as possible to complete the transition within the maintenance window. “This time, due to the scale of the patching, there is not enough patched capacity available to guarantee this,” Von Eicken wrote.