UC Berkeley In Brief
Main UC Berkeley website back online after early morning web server failure
BERKELEY – The main website at UC Berkeley, which went down at about 6 a.m. today (Tuesday, June 22), was back online by 9:15 thanks to prompt action by the server team and a fortuitous turn.
A team from Information Systems & Technology (IS&T) diagnosed the problem as a failure of the motherboard on the main campus web server, which is known as Arachne. Arachne is the home of the main campus website, the UC Berkeley NewsCenter, the campuswide calendar, and a number of other major websites.
As it turned out, the timing of the server failure could not have been better.
IS&T had recently acquired a new server as part of the preparation for the move of the main campus data center, located in the basement of Evans Hall for more than 30 years, to a new, state-of-the-art facility adjacent to campus. Acquisition of the new server was intended to help limit website downtime during the move and after that, to provide a degree of redundancy for the main server.
The migration of Arachne to the new data center had been scheduled for 9 p.m. Wednesday, June 23, when the main website was to go dark for about two hours. To expedite this transition, the new server had been set up to take over from the old machine. Over the past weeks, testing had taken place, the machine had been configured, and at about 5 p.m. Monday, June 21, files from Arachne had been copied onto the new server.
When the server failure occurred Tuesday morning, IS&T's Jann Fong (who manages the Arachne support group) consulted with owners of the websites hosted there and they jointly decided to accelerate the schedule to turn on the new machine, doing so immediately.
"The testing and preparation paid off," said Fong. "In effect, we just moved up the transition schedule. As a result, the scheduled outage on Wednesday night is no longer necessary."
People using the main UC Berkeley website may notice that files now load more quickly. "The new machine is zippier," said Fong, noting that the central processing units on the new server are significantly faster.
Meanwhile, the motherboard on the old server will be repaired. In the future, the two servers will give the campus a level of redundancy and limit web downtime.