Author: admin

Wed Oct 1 21:17:19 PDT 1997 — On October first we brought up a brand new T1 circuit direct to Sprintlink. This rounds out our direct connections, as we’ve now got direct fast links to the three largest backbones. This means low latency and packet loss, which translates to snappy, reliable Internet performance. This triad of circuits also means that our connectivity is sized so that an outage of any one circuit will not impact our network. Next time you’re shopping for ISPs, ask ‘are you connected directly to the big three?’

Wed Oct 1 21:09:39 PDT 1997 — We’ve had four major problems this evening. Busy signals were the first trouble – we’ve been adding capacity, but it seems that we’re not keeping up. We’ll be bringing more lines up tomorrow. The web and ftp servers were also offline for a brief time, and mail was a bit overloaded and slow. We’re scaling things up here, so we shouldn’t see more episodes like this. Sorry! -Dane

Wed Oct 1 20:34:47 PDT 1997 — We’re currently having problems with ftp.sonic.net and www.sonic.net. Both are currently being restarted right now, and should be back online shortly. -Dane

Mon Sep 29 19:14:55 PDT 1997 — We’ve had some problems this evening with interactions between deployed releases of Livingston’s ComOS, which runs on our dialup access servers (pm to pm15). This caused problems with our primary authentication server. Basically, too many requests were generated, and the access server wouldn’t wait for a delayed response, but would generate yet more requests. This resulted in a snowballing situation where the authentication server could not keep up. Our Computone equipment was willing to ask a secondary server for authentication, but the Livingston gear seemed to assume that if it could connect, it would get a response, and just kept asking for one.

We’ve upgraded the operating systems on all of these devices, and in the process, managed to boot lots of folks offline as they were rebooted. Sorry about that – if your connection got terminated tonight, that’s probably why. It’s for the greater good, as the new OS is more stable and shouldn’t have problems like this one. -Dane

Mon Sep 29 06:46:25 PDT 1997 — The popmail server at pop.sonic.net is back in operation. It stopped working just before 9pm last night. The pop server at sonic.net wasn’t affected, nor was outbound email. Our monitoring system didn’t see the problem. Currently, we watch the ‘inetd’ program on the pop server, which serves the pop connections for clients — unfortunately, inetd continued to operate correctly for everything but the popmail service. We will start monitoring the service directly. We apologize for the problem, and we’ll try to not let it happen again.

Mon Sep 22 23:31:22 PDT 1997 — I noticed sluggish performance on one of our terminal servers (pm14), and did a bit of packet dumping. It turns out that someone at osu.edu doesn’t like us, and was attacking our terminal server. I’ve filtered this host at our gateway, and have filed a complaint with odu.edu. I love Cisco IOS – it’s how the Internet works. -Dane (note correction – odu.edu, not osu.edu)

Mon Sep 22 22:42:18 PDT 1997 — One of our 18 network access servers went gone offline, and it just happened to be the first one. I’ve just rebooted it, and all’s well. Establishing a connection may have acted strangly for about 15 minutes. I’ve flashed up the OS, and at the next reboot it’ll be running the new version which should eliminate this type of crash. -Dane

Sat Sep 20 17:58:23 PDT 1997 — news.sonic.net was offline for a brief time this afternoon. The server software had some problems and exited. -Dane

Thu Sep 18 14:35:31 PDT 1997 — Pacific Bell is just about to end an installation discounting program for 128k and above Frame Relay dedicated circuits. If you are thinking about a full time connection, you can save about $700 by ordering before September 26th for installation before December 31st. If you’re interested, please email ec@sonic.net or call 522-1000 x201 for details. -Dane

Wed Sep 17 09:49:13 PDT 1997 — The very first modem in our X2 hunt group had problems this morning, and this seems to have affected the entire group. X2 was unavailable for about 45 minutes. We are currently programming a replacement modem, and X2 will be offline for about five minutes while we put that into service at about 11am. The modems are 16 port USR Total Control rackmount chassis, so when one port fails, you’ve got to replace the whole thing. -Dane and Brian

Update – the replacement unit is acting a bit oddly, and Brian is working on getting it to perform correctly. Sorry for the intermittant availability of X2 this AM. -Dane