Feeds

Google's Postini Fail pinned on bad filter, hardware glitch

Oh, and 'malformed types of messages'

Secure remote control for conventional and virtual desktops

The extreme email delays the plagued users of Google's Postini message management service earlier this week were caused by a shoddy email-filter update and a power-related hardware failure involving the company's database storage servers.

Today, the Mountain View Chocolate Factory released an "incident report" to Postini users, saying the "severe mail flow issues" began at 11:30pm Pacific time on Monday and extended through at least 12:30am Pacific on Wednesday. That puts the email snafu past the 24 hour mark.

The report does not say how many users were affected. Google tells us the problem was limited to customers on Postini's "System 7," one of several systems running the hosted email security and spam-filtering service, but at least one customer says the problem extended to System 5 as well.

"My company is on System 5 and our email was pretty much non-existent until we switched to a backup system. Once we pulled Postini out of the loop, all of that deferred mail hit our system (along with quite a lot of spam)," said Russ Meyer of the US-based Midland Paper.

At one point, Google rerouted traffic to another data center, which could explain the delays seen by Meyer.

Unlike so many on System 7, however, Meyer and Midland never had problems visiting the service's web-based admin console, which Google switched off for some customers in an effort to boost mail flow.

On Monday evening, after Google's monitoring systems detected the problem, engineers rerouted mail traffic from what the company calls a secondary data center. But this didn't help. So they returned some of the traffic back to the primary facility "to maximize processing resources." Then, at least for some users, they shut-off the admin console and some other web interfaces in an effort to reduce the strain on those resources.

Eventually, Google engineers decided the problem was down to three things:

  • A new filter update appears to have inadvertently impacted the mail processing systems.
  • Unusual malformed types of messages triggered protracted scanning behavior, and its interaction with filter update affected mail delivery.
  • A power-related hardware failure with database storage servers reduced input/output rates. The latency in database access reduced our overall processing capacity.

Which sounds like two things to us. Surely, it's the service's duty to deal with "malformed types of messages" - whatever those are.

"The combination of these conditions resulted in high failure rates for mail processing and the deferral of new connections from sending mail servers," Google's report says.

On Tuesday evening, a day after the delays first hit, engineers replaced the faulty hardware - with help from the vendor - and at 11pm Pacific, Google says, database disk throughout returned to normal. Then, an hour later, Google removed the offending filter update, and according to company, mail processing was back on track.

Google continued to process traffic across both data centers for another hour. The company does say, however, that users may still experience delays. "Although mail processing was at normal speed and capacity, some users may have seen delayed messages continue to arrive in their inboxes. These potential delays occur when the initial or subsequent delivery attempt is deferred and the sending server waits up to 24 hours before resending the same message." This explains complaints we received on Wednesday afternoon.

The report says no messages were bounced or deleted.

Originally, Google indicated the problem was limited to US users, but yesterday, the company acknowledged that at least some European users were affected as well. ®

Security for virtualized datacentres

More from The Register

next story
UNIX greybeards threaten Debian fork over systemd plan
'Veteran Unix Admins' fear desktop emphasis is betraying open source
Netscape Navigator - the browser that started it all - turns 20
It was 20 years ago today, Marc Andreeesen taught the band to play
Redmond top man Satya Nadella: 'Microsoft LOVES Linux'
Open-source 'love' fairly runneth over at cloud event
Return of the Jedi – Apache reclaims web server crown
.london, .hamburg and .公司 - that's .com in Chinese - storm the web server charts
Chrome 38's new HTML tag support makes fatties FIT and SKINNIER
First browser to protect networks' bandwith using official spec
Admins! Never mind POODLE, there're NEW OpenSSL bugs to splat
Four new patches for open-source crypto libraries
prev story

Whitepapers

Forging a new future with identity relationship management
Learn about ForgeRock's next generation IRM platform and how it is designed to empower CEOS's and enterprises to engage with consumers.
Cloud and hybrid-cloud data protection for VMware
Learn how quick and easy it is to configure backups and perform restores for VMware environments.
Three 1TB solid state scorchers up for grabs
Big SSDs can be expensive but think big and think free because you could be the lucky winner of one of three 1TB Samsung SSD 840 EVO drives that we’re giving away worth over £300 apiece.
Reg Reader Research: SaaS based Email and Office Productivity Tools
Read this Reg reader report which provides advice and guidance for SMBs towards the use of SaaS based email and Office productivity tools.
Security for virtualized datacentres
Legacy security solutions are inefficient due to the architectural differences between physical and virtual environments.