Month: August 1999

Tue Aug 31 23:56:04 PDT 1999 — The migration of our PacBell DSL customers to our huge new T3 circuit went well. Downtime for 95% of the customers was 30 minutes, and the other 5% took about 15 more minutes to get configured and online. The new T3 offers a full 40 Megabits of bandwidth, as opposed to our old circuit which was only 1.5 Megabits. This means blazing speed for those bandwidth hungry DSL modems, and no congestion during prime time! In addition, the new router supports a secure arp method which allows DSL users to reach each other’s PCs directly, for online gaming for example.

Please post any feedback to news:sonic.net – thanks! -Dane

Tue Aug 31 16:43:44 PDT 1999 — Our final testing of the new DSL ATM T3 is complete, and it’s doing really well in our test setup. We will be making the transition to this circuit tonight at 11pm. Expected downtime is 30 minutes. Once complete, DSL customers will be on this new circuit which is about 25x as fast as the current circuit, and has lots of room for more new customers. This will also resolve a Cisco bridging issue whereby DSL users currently can’t directly reach other DSL user’s PCs. We’ll post a followup here after the transition to let you know how it went. -Dane

Mon Aug 30 16:15:43 PDT 1999 — We contacted USR today regarding disconnect issues with the 522-1003 dialup pool. They took a look at our gear and made some suggestions. We have made a config change on that gear which should help stop the disconnects. If you continue to have problems please post a message in the sonic.help news group. Thanks – Steve

Fri Aug 27 17:45:32 PDT 1999 — This afternoon, I had our core router in debug mode (watching BGP updates) and for some reason it caused the Cisco to crash. This was entirely operator error (mine) and I had to reboot the Cisco. Total downtime was 6 minutes, and I apologize for the error. -A mortified Eli

Thu Aug 26 12:36:55 PDT 1999 — The shell server just had the same problem as the web server. We suspect someone is developing a nasty cgi that is consuming resources…we have increased the resources available on Bolt, the shell server, which should help prevent a future reoccurance. -Dane, Dom, Eli, Ian, Scott

Thu Aug 26 14:55:42 PDT 1999 — Mail server upgrades. We have shifted mail traffic to a new mail server, Prop.sonic.net. Prop is a clone of Sub, the old mail server, which means it is _not_ running qmail. We are monitoring the new host, and expect this transition to be much more seemless than our last try at upgrading. -Scott and Eli

Tue Aug 24 09:53:30 PDT 1999 — The web server that serves our multihomes (Thunder) required another reboot this morning. The error message was ‘No free inodes – Contact Linus’, which I will not undertake until we’ve checked some other things first. Downtime was about 8 minutes, and we will continue to investigate the issue. – Eli and DaveJ

Tue Aug 24 09:02:10 PDT 1999 — After recieving complaints that some people could not see web sites beyond sonic.net, I did some investigation and found one of our terminal servers had lost its default route. I added it back in and all is well. We apologize for any frustration this may have caused. – Steve

Mon Aug 23 21:56:29 PDT 1999 — After much fuss with PacBell, we’ve got the downed PRI circuits back online. A correction on the previous MOTD – this trouble took offline 25% of the capacity in just one hunt group, the 1003 group. This is just 184 phone lines, _not- 25% of our total capacity. We did return some busy signals in that group this evening, as it took PacBell until 9:30pm to get the circuits back online. -Dane

Mon Aug 23 21:50:23 PDT 1999 — We’ve moved back to a two server mail config to address performance issues. We’ll post more info here when we’ve got it. -Dane