Yesterday was just one of those reverse Midas touch days where every piece of technology I touched, so matter how trivial, seemed to up in flames. It would have been advisable to close the laptop, grab a book, and go outside to sit under a tree… but it was 115 degrees in the shade yesterday, so there went that option.
It started with some loud, metallic clicking fan-like noise emanating from the left side of my MacBookPro. This was more noisy than the “hum” or “mooing” noises often reported, and it grew in intensity as I hit my most heavy CPU usage app, the Second Life client. On restart, it would settle down to a lower steady noise, but it does not sound healthy. First step, run a backup on everything. Check the Apple Support forums. Nothing there comparable. Run the extensive hardware tests. All passed. So this means, I have to drive it over today to the Apple Store, which is not a pleasure trip to browse the cool new toys, and face the fact I could be without my work machine 10+ days if it gets shipped out.
Then, I thought I was making a very minor change to our Apache configuration for some new virtual domains on the NMC site. I have experiences and a bit of awareness of just editing the raw httpd files, have done it on several servers and on several copies of apache I run domains on my laptop (useful for testing web apps). It is really not all that complicated for name-based virtual domains. But out site has this WebMin interface, which is handy for some tasks, but there are about a million screens under Apache, and it did take a bit of banging the first time to set up a new virtual domain. This time I merely had to set up a new one that would point to an existing web content directory, as we do a shuffle for a domain that needs to be renamed/
Something went south, as now all virtual domains pointed to the main web site. I deleted them all and rebuilt them as I thought I had done before. I read through every piece of Apache and WebMin documentation I could scrounge. No dice. The problem is that the newest domain is really a media source for a new streaming service, and it is the one our content delivery system hits as it caches the content out on its system… so instead of caching the audio/video content I had set up, it was caching the entire web site.
A call to our co-located web support was a circus. After being on hold (“we are experiencing large call volume…”), explaining the situation to the human who finally answered, I was put on hold– I thought he was off to trouble shoot our configuration. It was about 15 minutes of hold time when the phone line just went dead.
The next sounds coming form my office are not repeatable.
On the next call, while waiting for the handoff from the phone tree, it hung up again.
More un-repeatable sounds.
On the third call, I got someone right away. Bad for him. After a short time on hold, he reported that the first person I talked to was not pecking at the command line, but “speaking with a supervisor.” And here is the deal- they are our webhost, set up this WebMin interface, but offer no support for it because we are a “1U” site… which I don’t even know what it means, but that they were not going to offer any help.
And here is the killer, I asked where can I get some help– he replies, “well, you can always post a question to a newsgroup.”
N E W S G R O U P ?????
What decade are we in?
Many more un-repeatable sounds.
I spent another hour going through things, checking error logs, got a former colleague who doe server admin chat with me as we tried to sort through the settings (thanks Derek), and finally, it was just time (an hour later than dinner) to just quit, and watch some brainless TV.
And as I laid in bed, it just gnawed away because it was obvious there was something so trivial it was invisible. Sleep was not happening, so I did some more googling on virtual host and vhost.conf (configuration files). Finally, on one of the Apache doc sites, I found:
If you are trying to debug your virtual host configuration, you may find the
Apache -S command line switch useful. That is, type the following command:
I had to dig around because our httpd was sitting elsewhere, but found the right clue:
VirtualHost xx.xxx.xx.xx:0 overlaps with VirtualHost xx.xxx.xx.xx:0, the first has
precedence, perhaps you need a NameVirtualHost directive
which did not sound right, since I had a
NameVirtualHost xx.xxx.xx.xx directive sitting in the main httpd.conf file. But what the heck, I through an extra copy in the top of the vhosts.conf file, tickled Apache (to restart the server proces), and SHAZAM! Life was good.
I was on such a roll now I dug into another issue with getting the right media playback address for our audio streaming service we are going to use to bring live audio into Second Life…. I had stumbled with *.m3u and *.pls sites to get the right alias for our streaming address, and just going the one simpler route, and using the provided address (which we were told “not to use in a browser”, but this was not in a browser), and SHAZAM #2! Another tiny task that feels like taking the Yellow Jersey for 5 days in a row.
All of this means nothing to anyone, and I cannot imagine why anyone would even be reading this. It’s my “diary”, eh?
Hoping the tech roll starts on the SHAZAM side today.