Amazon Web Services Inc. (AWS), announced that it will reboot Elastic Compute Cloud instances (EC2) across all regions and availability areas on Thursday evening.
After privately emailing affected customers on Wednesday, AWS detailed the reboot in a blog posting. AWS stated that the planned reboot was a “timely operational and security update” in response to a forthcoming Xen security update. The company will not release more information until Wednesday, after the Xen security reports are made public.
According to the AWS blog, “These updates must be completed before the issue is made publicly as part of an upcoming Xen Security Announcement(XSA)”. The details of this update will be kept confidential until then, in accordance with security best practices. This notice addresses many Xen environments and is not limited to AWS.
AWS stated that the reboot is not related with the “Bash” bug, which was discovered Wednesday by researchers.
According to RightScale Inc.’s FAQ, the reboot will begin on Sept. 26th at 2:00 UTC/GMT (Sept. 25, at 7:00 p.m. PT) and end on September 30th at 23:59 UTC/GMT. (Sept. 30, at 4:59 PM PDT).
AWS stated that less than 10% of its EC2 fleet would be affected.
AWS stated that the instances that require the update will require a system reboot of the underlying hardware. The host will be unavailable for a few moments while the patches are being applied. While most software updates can be applied without a reboot of the host, some updates require a restart. Reboots will be performed in a staggered manner so that no two availability zones or regions are affected at the same moment. All data and automated configuration will remain intact. Reboots should not cause any major problems for most customers.
One customer of AWS commented on the company’s forums that AWS had warned affected customers they wouldn’t be able avoid reboots by relaunching instances. AWS moderator clarified that customers can stop and start instances, but may not be able get a patched host.
RightScale says that the reboot will not affect T1, M2, R3, or HS1 instances EC2, but it will not affect T2, M2, and R3. RightScale cautioned that other AWS services, including RedShift, ElastiCache and RedShift, may experience downtime during the reboot period.