Peep
I've installed peep, and I must admit, it's rather cool.

Now a variety of different birds chirp away when we see web-traffic, and the species of bird depends upon what website is being viewed. Frogs croak when we see bounced mail, birds warble when legitimate mail is received, and doves coo when we receive spam.

As the CPU load increases the faint tinkling of a brook increases into the roar of a river, the sound of crickets is used to indicate the strength of inbound traffic, and the sighing of the wind for outbound traffic.

A goose can be heard in the distance in the case of changed disk SMART reporting, and a rooster crows (quite noticably!) in the case of a more serious error.

The surprising thing is, this isn't distracting. It instead forms a nice calming background of sounds. In fact, it's much more disturbing when I don't hear anything. It certainly makes a nice change to the background of whirring fans.

Writing new clients for peep also provides to be quite straightforward, although the documentation on many of the options could certainly be improved. I also tweaked the server's code in a few places, so it no longer wired-in that it binds to all interfaces on a machine, or broadcasts its presence.

One thing that I do need are more sounds. I may go looking for them should I ever find myself with a spare moment.

After some pain, and many hours, the new box is happily chugging away. I took advantage of the downtime not only to upgrade the motherboard, but also the CPU and RAM, and also moving from a 2.4.18 kernel to 2.6.7.

One thing that was immediately obvious is that the 2.6.7 kernel is fast. The speed at which the machine started when I loaded the new kernel was almost jaw-dropping. However probably the most plasing thing is the speed of the hard-drives.

I'm watching our RAID rebuild itself at a swift 28M/second, which is a 650% the speed at which it used to run. At least I hope it's rebuilding -- it could be going really fast because it's just pulling sectors off the disks and throwing them into the bit-bucket.

Some things in the upgrade didn't work quite as well. The onboard gigabit ethernet does come with Linux drivers, but they cause a kernel panic whenever they're unloaded, or when the interface is brought down. That's not something I'm going to put anywhre near a production machine.

The onboard 'firmware' RAID also came with Linux drivers (iteraid), which didn't seem to have made it onto the CD, but could be downloaded from the manufacturers website. These are GPLed, and give access to the firmware RAID arrays as emulated SCSI drives.

At least, that's what they do for everyone else. I spent more time than I care to admit tweaking, loading, unloading, and reading the source-code for these drivers trying to get them working. They detect the hardware fine. They cheerfully report the drives or arrays I have connected. They just don't do anything more. I haven't yet got the 2.6.x drivers running yet. As such, of the four PATA channels in the machine, I can only use two. Very frustrating.

At least I can use the on-board sound. Maybe I'll get peep running after all.

If you're making a motherboard, and you're going to provide onboard IDE RAID, that's great, even if the product isn't fantastic. However, should the user specify that they would just prefer to have some IDE channels, please do so in a standard fashion.

I shouldn't be up at 3am building kernel modules for something that should work out of the box.

As expected, as soon as jarich called me to say that she's obtained the replacement parts for the flaky server, all symptoms and signs that there were any problems ceased.

That won't stop me from replacing its guts anyway.

Australia gets the DMCA
Last night, the US-Australia Free Trade Agreement went through. Along with many other things, this will include "harmonization" with US intellectual property laws, including trademarks, copyright, and patents.

Somehow I fail to see this being of benefit to Australia.

Sick hardware
A rather important machine of ours went through a number of unexplained reboots a number of months ago. We thought we had diagnosed it as a problem with the UPS, as removing it from the system resulted in sable operation.

Last night, the same machine experienced another set of reboots without any apparent reason. However this morning I discovered the culprit, after disabling the watchdog that was running. It appears that the IDE controller (or the kernel, talking to the IDE controller) completely dies, with many 'lost interrupt' messages on the console for three of the four drives in the machine.

Looks like I'll need to schedule downtime and replace its guts. Dealing with hardware is probably my second least favourite of all sysadmin tasks. (Restoring from tape is my least favourite)