Mail and Network Outage, 8/23/2008 & 8/24/2008

Executive summary for those who don’t want to read anything beyond the bare details: On Saturday, August 23rd, both the MCS Mail & Calendar (aka Zimbra) and the MCS Production network will be undergoing maintenance and will be unavailable from 6AM until 6PM. Also, on Sunday, August 24th, Zimbra will be offline from 8:30 to 10:30 AM.For details, read on.I’m looking at the weather forecast for Saturday.  I find it interesting that the weather is one of those things about which you can convey a great deal of information using nothing more than a simple drawing.  Big smiley sun — a sunny day.  Angry cloud blowing icicles — cold and windy.  Giant turtle wearing a wizard hat and eating skyscrapers — stay indoors.I’m looking at Saturday’s forecast because I was hoping I could be the happy weather dude and tell you all it was too nice a day to be working on your computers and you should go outside.  But, alas, the little picture is of a gray cloud with lightning bolts.  And lightning bolts are almost never good.  Unless you’re Thor, in which case they’re how you earn your paycheck.But, I digress.  Regardless of what the weather’s doing, you should not be trying to use MCS core computing services.  I’ll explain all the details, but first, let’s try the weather approach and see if we can’t capture it in a drawing.

Click to enlarge

Yeah.  That about captures it.Okay, the details, for those who can’t interpret the above:The Zimbra server is going to be upgraded to the latest version (5.0.8) on Saturday.  CIS will send a detailed announcement of this to all Zimbra users across the laboratory, but I wanted to make sure you were all aware of this.  During the outage, no mail will be sent or received, however no mail will be lost, only delayed.After the Zimbra upgrade is complete, we’ll have preliminary iPhone ActiveSync support.  I’ll send detailed instructions on how to take advantage of that once it’s done.  There are other improvements coming in 5.0.9, but that’s still not released.Also, the disk that the Zimbra server is using will be undergoing a firmware upgrade on Sunday, 8/24/2008 from 8:30 until 10:30 AM. During this period, the Zimbra service will be offline with the same ramifications as above. Expect more details from CIS on this.That part of the maintenance weekend is pretty trivial.  However, we’re also doing a major network shuffle during this maintenance window that’s going to make parts of the divisional computing infrastructure unreachable for various parts of the day.Specifically, for security and functionality reasons, we need to split apart the network we’ve always internally called “BorgNet”.  It’s the network that spans 140.221.8.0 through 140.221.11.254.  The work will be starting at 6AM and will proceed throughout the day.  Some machines will be off the net for most of this outage window, mainly desktops.  Critical servers will have their downtime minimized as much as possible.  Due to the nature of the network split, some hosts will have new network addresses after the move.While we expect the work to take less time than allotted, we’re going to give the outage window as Saturday, August 23rd, from 6AM until 6PM.  Any connectivity you experience during that window is coincidental and is not to be construed as an all-clear.  In fact, the only thing that should be construed as an all-clear is a message from someone in Systems indicating the work is finished.We do not anticipate this affecting connectivity to the ALCF big machines, or the compute clusters in MCS.  The wireless infrastructure will likewise be unaffected. The web server will have a short outage while its network is reconfigured.  Anything that uses our divisional database servers will likewise have a short window where connectivity is out.  As stated above, we’re going to keep those outages to a minimum.If this work poses an undue hardship, please let us know and we’ll see what we can do. We will send a reminder of this work on Friday.Thanks!