Two activities impacted service during this time frame. First; we had to install new libraries on all of the machines running SunOS, requiring a reboot. These addressed buffer overflow and apparently some Y2K concerns. Second; Sprint is testing one of our T1's. Since they did some work on route reflectors back on October 8th, we have been seeing packet loss from between about 2% and 10%, averaging about 4-5% on both links. They are seeing some CRC errors on ONE link though it amounts to about one CRC error a day; hardly enough to account for the packet loss; but they want to eliminate the circuit problem before looking at other things. The fact that it happens on both links and the other link isn't taking CRC errors apparently is not a sufficient clue. Additionally, this testing was schedualed for 2am today and they were supposed to call me FIRST so that we could route all traffic to the link that would be staying up (one link being sufficient to carry the load during this time frame), but instead they took the link out 15 minutes early without calling first. We did re-route traffic as soon as I became aware that they had taken the one link out of service but it did cause about a ten minute interruption in network connectivity.