Priority Colo Inc.

Hi guys,

This notice is to alert customers who are either on the shared hosting platform (www2 through www6), and all colocation/dedicated customers on the sw01.prioritycolo.com switch, that we will be performing maintenance on Sunday August 7th, 2005, between the hours of 11:00am EST, and 4:00pm EST. We will be upgrading the daughter cards on both the primary, and reundant/secondary supervisor engines, from 100mbps FX fiber, to 1000mbps Gig-E fiber.

Presently, sw01 has 4 x 100mbps links connecting it to msfc02, and 3 x 100mbps links connecting it to msfc01, to provide both physical and logical redundancy for the customers attached to the switch in question. Once this upgrade is complete, we will have 2 x 1000mbps links connecting sw01 and msfc01, and 2 x 1000mbps links connecting sw01 and msfc02. This will be performed to improve redundancy, and potential throughput between the devices in question.

This maintenance window will involve between 2 and 4 short outages for customers on the sw01 switch, as the supervisor will have to be "failed" between the primary, and secondary supervisor units, which takes a couple of minutes each time, so the cards can be pulled, physically upgraded, and reseated as necessary.

Customers who are not on the sw01.prioritycolo.com switch will not be impacted by this maintenance.

If you have any questions, please direct them to admin@prioritycolo.com

Regards,

This maintenance has been sucessfully completed (as of ~3pm), though some customers were down longer then intended. Upon the initial failover/reload, sw01 lost its configuration for blade 8 (unknown reasons), and this was not realised until all of the testing on both new cards was completed, where it was remedied by restoring the blade 8 configs from last nights backups.

Customers on blade 8 (including our www4 server) noted a single ~30 minute outage, instead of the one 3 minute, and one 4 minute outage (~25 minutes apart) that the rest of the customers should have noted. We're presently investigating what caused this, while this did not present to be a critical issue (no errors logged, etc.), it was indeed a nuisance, and most likely a cisco bug (as only 1 of the 12 currently utilized blades, or 1 of 9 identical blades on sw01 was affected).

Regards,