DNS Server and ISP Troubles

I'm beginning to see why people sometimes want to run their own DNS (Domain Name System) servers. My SBC DSL's DNS servers keep flaking out from time to time, more often lately. So when I'm surfing the Web it suddenly starts telling me that the site I'm trying to visit, such as Slashdot.org or Wireball.com, could not be found, and when I try to ping them, they report "unknown host ...".

I got sick of this today, so I reconfigured my Internet connection from SBC DSL's auto-configured Los Angeles DNS servers to a couple of OpenNIC's Public DNS Servers (one in New Orleans, LA and one in Tokyo, Japan), and since OpenNIC doesn't recognize .biz domain names (I gather), I put down SBC/Pacbell's Houston, TX primary DNS server as my third static DNS server. Using OpenNIC servers also lets me access such domains as .glue, which I could not access before because my SBC DNS servers do not support OpenDNS.

Of course, I could have tested it by trying to Telnet into my ISP's DNS servers on port 53, as detailed in this article. I did try it on the OpenNIC DNS servers, and it turns out that the San Francisco California and Phoenix Arizona servers were not responding at the time I did the test.

So far it seems to be working well - it's quite snappy, in fact. At first I had configured it without checking to see if the DNS servers were up, and it was taking a 2-4 seconds to resolve hostnames; I checked the DNS servers and it turned out that the first one I'd entered (in San Francisco) was not responding. Fixing that speeded it up greatly. Nice to have it working and not claiming that common sites do not exist (and since I'm using wildly different servers in different geographical regions, they're not as likely to go down all at the same time, unlike SBC Pacbell's local DNS servers).

From what little I know about DNS servers, it sounds like running one's own software/Linux- or FreeBSD-based DNS server (or cache/proxy) would be nice - you would know and be able to see what was going on, you could request name resolution from multiple servers if the first one failed to respond promptly, and have many more than just two or three DNS servers to choose from, so you weren't as subject to the whims of flaky servers, and even do statistical analysis on the different servers so that it tended to request name lookups from the fastest and most reliable servers more often. At least, I assume that's all possible. Yet another reason for me to learn Linux more thoroughly.

91 Honda Accord Maintenance - Keeping It Running

I thought it might be interesting to write a bit about my used '91 Honda Accord LX 4-door sedan, and what I have to do to keep it running. Some of the maintenance info may be useful for those of you with similar vehicles or problems. Even if not, it may be interesting from an anecdotal standpoint.

So this Wednesday, I got a phone call from one of my family members - the car had refused to start in the mail service parking lot, and since I have some technical aptitude with cars as well as computers, they wanted to know if I could head over there and see if I could get it started. I hemmed and hawed a bit, because it was three (3) miles away and I wasn't used to walking quite that far, but in the end I stuck some tools in my pack, and headed over there.

It sounded like it might be a repeat of a problem we've had a number of times with this vehicle, and which is common with a lot of Hondas, including relatively new Hondas such as the CR-V. The vehicle turns over when starting, catches, and then as soon as you release the key, dies. This is caused by dirty contacts in the ignition keyswitch. The tools I brought were primarily a pair of pliers (to disconnect the battery), philips screwdriver (to get the steering column cover off, and unscrew the ignition switch), a piece of sandpaper (for cleaning the contacts), and silicone grease (for protecting the contacts after I was done sanding the corrosion off).

When I got there, the car wasn't there, and I couldn't find the family member at the agreed-upon spot. I finally headed back another three miles to home, and when I got there, the car wasn't there either. There was, however, a message on the answering machine - the family member had gotten the car started, headed home with the groceries, and then after dropping the groceries off had driven over to the second-nearest Arco gas station (which was a mile away). And now it was stuck at the gas station and wouldn't start. Nggh!

When I got to the gas station, the car was there, and not only would it not start, it hardly wanted to turn over. All the indicator lights on the dashboard lit up brightly, and the headlights were plenty bright; however, when I tried to start it the indicator lights would practically black out (not merely dim), and the engine would only move about half a revolution. Figures I forgot to bring my tools this time. So we pushed it over by the curb, because it was getting dark, and decided to come back in the morning.

In the morning, we brought an electronic multimeter with the rest of the tools, and measured the voltage both while the key was in the ignition and while trying to start it. Merely sitting with the dashboard indicators lit up, the battery measured 11.7V - a bit under the 12.8V it normally reads. However, when I tried to turn the engine over, the battery voltage dropped to just 2V! The battery at this point was over three years old, and past its warranty, so we figured it was a safe bet that the battery was probably at the end of its lifespan.

We didn't want to pay for a tow or a taxi, and Sears was unfortunately 2 miles away, so we thought we were going to have to lug a 50-pound or more battery 2 miles back from Sears on foot. Luckily, it turned out that Orchard Supply Hardware, which was just a couple of blocks away, sold car batteries, so we got a nice 12V with a 1-year warranty for about $74USD after tax, and carried it back between the two of us (switching arms once). We measured the voltage (over 12.8V), installed it in the car, and it fired right up!

In the weeks prior to this, the car had been taking more and more cranking to get started, and I thought it seemed like it wasn't cranking as fast as it used to. Well, with the new battery, it was cranking at least half again as fast as with the previous battery, and it caught in much less than a second, compared to the old battery at around 2-3 seconds. Also, the voltage dip while starting it was pretty small - it dropped to about 11.5V momentarily while cranking the high-amperage starter, and then jumped up to the 14.5V alternator voltage.

So, there you have it. Not a serious problem after all; just a dead battery. Seems pretty obvious in retrospect, but we had been getting pretty worried about the age of the components in the car; a lot of them are well past their recommended service interval, and it just seemed natural to assume the worst :) Next week: how the car behaves when the alternator is malfunctioning, and how we fixed it ourselves for $25 (hint: the brushes were just worn out).