Sunday, March 19, 2006

The "new" AT&T - but who do I call?

I loved SpeakEasy at our old place, great support, fast connection, no issues. When we moved into this place the only affordable "broadband" option available was IDSL (144/144) via Covad's Business unit. San Jose's cable network was still owned by AT&T broadband at the time and not comcastic. Not even PacBell/SBC offered DSL in this southern end of San Jose. So we ordered IDSL and got it installed, since going back to modem was just too painful. The installation basically put ISDN on the second phone line and combined the 3 IDSN channels into a single 144kBit pipe. Not great, but bearable.

A year later we got a call from SpeakEasy offering free upgrade to 1.5/768 ADSL using a "remote terminal" SBC had installed in our neighborhood. Not only was this faster, but even cheaper. I didn't have to do the PPPoE nonsense, could continue to use my two static IPs and run servers from home. That connection rides on the regular SBC phone line like most other ADSL lines, but uses SpeakEasy's backhaul from the central office. So far, so good.

Fast forward to a few weeks ago.

DSL is down. Phone is dead. Patricia calls me, I tell her to find the customer service number on the last phone bill, report the problem and have someone come out. She did, a technician showed up the next day, tested the line, and eventually found the culprit: One of the connectors had slipped off. Funny how that happened only a few hours after Patricia saw someone from SBC working on the phone closet in the neighborhood.

Fast forward again. SBC is now the "new" AT&T.

DSL is down. Phone is dead. I can't listen to my favorite Internet radio station. I pull out the last phone bill and look for customer service. Hmmm. Nothing on the front. Only, very prominently,
Lots of fine print on the back. Phone numbers of the PUC, the FCC, complaints addresses, but no phone number at AT&T to call in case of a problem. This is the phone company!

Finally, I go back to an older SBC bill, look on the back and indeed they have a customer service number in the fine print: 1-800-310-2355 (i.e. 1-800-PAC-BELL).
Call em up on my cell phone. "We're sorry, our offices are now closed". At least they let me connect to their 24 hour repair service. I get the ticket in. The whole process with their automated system takes "only" 15 minutes. "The next available time slot is Tuesday, March 21st" . Arrrgh. That's two days from now! They expect me to be without phone service for two days? I'm angry and annoyed.

Magically, everything starts working again just fine early in the afternoon. I'm guessing someone came out and slipped the connector back on...

Saturday, March 11, 2006

Wow! Dust and temperature...

chef was running in its new location behind the CD shelf since end of December. Yesterday evening I took it out to look at the fan configuration. Boy, that box was DUSTY!
Removed dust coating from hard drives, and cleaned the fins of the CPU heatsink. The impact is pretty impressive:

Chef disks temperature

Chef case temperature

Chef fan speeds

Note, how the CPU fan speeds up after the cleaning, as well as how the temperature of the hard drives drops as there is no more dust to insulate them from the cooling wind of the fan mounted in front of them.

Sunday, March 05, 2006


So, I finally ran into my first rejected mail due to SPF:

-Queue ID- --Size-- ----Arrival Time---- -Sender/Recipient-------
B79A65361 1886 Sat Mar 4 16:20:14
(host[] said: 550-5.7.1 {mx020} The recipient does not
accept mails from '' over foreign mailservers.
550-5.7.1 According to the domain's SPF record your host '' is not
a designated sender. 550 5.7.1 ( ) (in reply to RCPT TO command)) is still hosted on SpeakEasy's DNS servers (I didn't get around yet to change that to and run secondary on ns1/
SpeakEasy doesn't expose SPF records to their DNS customers. According to the SPF record

v=spf1 ip4: ip4: ip4: ip4: mx ptr ~all

as published by SpeakEasy results in a "softfail" condition for SPF. In other words GMX refuses mail on softfail. Niiiice...

The remedy is easy:
Add relayhost=[] in /etc/postfix/, restart postfix, done.