At 1037 EST, 1537 UTC, the authentication server for the mail storage servers stopped responding to authentication requests. Unable to determine the reason for failure we restarted the server at 1055 EST, 1555 UTC, and it is now functioning properly. We will continue to investigate the cause of this failure.

Update on the Jan 26, 2005 power outage.

Batteries for the failed UPS system arrived by truck from Tampa and were installed late yesterday afternoon. By 1830 EST, 2330 UTC, we were back on UPS power.

Power Outage Update: January 27 0330 EST, 0830 UTC

At 1505 EST, 2005 UTC, a UPS in the data center failed and our systems were without power for approximately 20 minutes. The engineers at the data center attempted to bring the UPS back online without success and utility power was switched to the circuits fed by the failed UPS. When power was restored one router failed to boot and its HSRP backup failed to announce our routes contributing to the problem.

All servers suffered some file system damage requiring another 10 to 90 minutes to repair before the servers could be brought online. Approximately 80 minutes after the power failure the incoming MX servers, the outbound SMTP servers, the MX relay servers, and two mail storage servers were back in service.

The last mail storage server was back online just before 1800 EST, 2300 UTC. The web sites were brought up shortly after all mailbox servers were up and functioning. The Account Manager followed shortly after that.

The UPS failure was not an inverter failure as reported earlier. When the Powerware engineers arrived on site diagnostics showed that the UPS was healthy but the batteries were not up to the task. Throughout today the facility was experiencing short but serve utility power fluctuations causing all of the UPS systems to switch in and out of battery operation and the batteries were not being fully recharged when returning to utility power. This frequent discharge and inadequate recharge caused the batteries in this UPS to fail resulting in this outage.

The data center is expediting delivery of new batteries for this UPS and the other UPS systems in the data center. The new batteries may arrive as early as tomorrow afternoon, Friday.

We sincerely regret the inconvenience to our customers caused by this unfortunate outage.

Due to the impending arrival of hurricane Frances in South Florida, customer service will not be available by phone or email on Sep 3, Sep 4, and possibly Sep 5 if Frances takes a more Westerly direction.

There is no concern that email services will be impacted by hurricane Frances.

Squirrelmail support for composing a message in HTML format was added earlier this month. HTML compose can be enabled on the 'Options -> Display Preferences' page. The selection is near the bottom of the page. HTML compose will work with recent versions of Netscape, Mozilla, Firefox, and Internet Explorer.

A module has been added to Squirrelmail that can manage IMAP folder access lists. IMAP access lists allow you to share one or more of your mailbox folders with others.

Direct access to Sieve scripts is now permitted by connecting to mail.mxes.net port 2000 using the MANAGSIEVE protocol. Mulberry is the only mail client we know of that can manage Sieve scripts.

There were several periods of routing instability during our data center's scheduled maintenance window this morning. The instability occurred between 0245 and 0350 EDT (0645 and 0750 UTC).

June 5 between 1600 and 2000 EDT (2000 and 2400 UTC) we will be testing fail over of all routers, switches, and server interfaces in our network. During this testing period there will be periods of less than 1 minute where some or all servers will not be reachable.

We apologize in advance for any inconvenience caused by this testing.

An unscheduled outage occurred from 1535 EDT to 1645 EDT (1935 - 2045 UTC). A partial failure of a Ethernet line card interrupted connectivity to our primary providers but the failure mode did not cause the backup router to assume routing duties. The line card has been replaced and the exact nature of this failure is being investigated.

We apologize for the inconvenience caused by this outage

Per mailbox Bayesian classification has been added to all qualifying accounts. The Bayesian classifier is not enabled by default and it must be trained before it will function. The classifier configuration is located under the 'Manage Mailboxes' menu selection.

We experienced an unscheduled outage from 1515 EDT to 1625 EDT 2015 UTC). This outage was due to an OC-12 line card failure in an upstream switch and intermittent connectivity problems via backup routes.

We apologize for the inconvenience caused by this outage.

Later this month per mailbox trainable Bayesian classification will be available. To accommodate this new capability modifications to the Account Manager have been made that allow an 'Access Policy' to be enabled to provide Allow/Deny lists without having to enable spam scoring on that policy. To reflect this change in functionality, what was a 'Spam Scoring Policy' is now called an 'Access Policy'.

This change does not impact any existing 'Spam Scoring Policies' other than a name change and adding a check box to disable spam scoring if desired. When the classifier is available you can choose to use it by itself or in conjunction with the existing spam scoring system to identify spam.

xbl.spamhaus.org has been added as an additional data source for the Open Relays and Proxies MX restriction. The XBL lists a lot of compromised machines that are currently spewing spam but are not detected as open proxies.