April 22, 2008 – nuxx.net

I just hit submit at Newegg.com to complete the purchase of a Tyan Transport GX28 server with a Tyan Thunder K8S Pro (S2882). As item number N82E16856152008 the barebones server was only $409.99, with $31.24 shipping. Thanks to a friend helping me out with some RAM and some disks which I’ll be able to make redundant I’ll likely end up with a 1U server with a pair of Opteron 800-series CPUs, 8GB of RAM, and mirrored 500GB SATA disks.

This all brings about an interesting question of where to host the new box. I’ve discussed this before, and the more I think about it the more I want to get my box out of Waveform. Things are working fine for now, but I question what will happen if or when the box does start to have problems. The provider I’m most seriously looking at would run $100/mo

So, now I just have to wait for things to be shipped and delivered and then I can start assembling it all. I imagine I’ll let it cook for a few weeks to a month before installing it. It’ll be running FreeBSD 7.0, likely with a custom kernel and world rebuilt specifically for the CPU. I think I’ll also want to give the new ULE scheduler a go, particularly after seeing this presentation (PDF) about where FreeBSD is going.

I’ll continue to stick with lighttpd, although I hope that the OpenSSL bug in 1.4.19 is fixed in ports soon. Disks will likely be mirrored with gmirror, although I will investigate the on-board hardware RAID. I’ll probably also stick with MySQL for the db and Postfix for mail. Basically, nothing will change in that regard.

I may opt to eliminate some individuals I currently host from the box, mostly because I never have contact with them. I don’t mind hosting people, but when the sites sit mostly unused and I have almost no contact with the individuals who use them (except when there are problems, of course), it’s a bit frustrating to keep up maintenance on apps running on the sites. Also, this new provider has stricter limits on bandwidth (1mb, 95^th percentile), and I need to be a bit more careful about how it is used. Anyway, if I’m opting to remove your site from hosting I’ll contact you outside of here and provide you with a chance to get your data.

For now I wait, then build. This could be pretty nifty. Oh, and the colocation provider offers IPv6 at no extra charge, so that ought to be fun to play with as well.

Leave a Comment

Two lines of display on the failing PLED during bootloader update. Note that the second line is more faint than the first, pixels are smaller than expected, and there is distinct fading around the edges of the display.

Last night I made quite a bit of progress on Ivan’s P3, getting it up and running, but I did run into a few problems. First, as can be seen above, the PLED which came with the enclosure seems to be failing. It will either display no text, one line of text, or (rarely) both lines, and every time pixels seem small and dim, with the edges of the display fading to nothing. Only a power cycle of the P3 seems to (re-)activate non-working parts of the display. As the P3 still functions even when nothing appears on the display, I believe that it’s actually the display elements of the PLED which is failing.

Here is the PLED in my P3 from 2006, and when it is compared with these two images (1, 2) of the PLED from Ivan’s P3, it seems pretty obvious that something is wrong.

Thankfully there was a spare PLED in the package of parts I received, so tonight I’m going to try that one instead.

Second, I’m not happy with the cables I made for connecting the tempo and data pots to the mainboard, so I’m going to remake them. The tempo pot connection seems to be a bit flaky, so I’m not sure the pins are properly seated in the connector. The cables are also short enough that they are difficult to connect, so I’ll probably redo them with braided 24 gauge hookup wire or something like that.

Finally, the v1.5 mainboard construction notes indicate that when the P3 is being set up for MIDI sync, D2 and D3 should be replaced with 220pF caps in order to add a bit of extra capacitance to the lines to work around false triggers. I didn’t have any 220pF parts, and none came with the kit, so I instead used 330pF parts. I don’t believe this will be a problem, but I made a post to the analogue-sequencer group asking for confirmation.

Beyond those three problems, the build is going quite well. I’m very happy with how both the step board and keypad came out. The IDC ribbon cables, their connection, and power input stuff also worked out great.

I was also able to get the latest bootloader and v3 firmware on the P3, which means that as soon as the other issues are sorted out I can get the MemX and latest v4 beta installed. After that it’ll just be time to install the knobs, test it out, and ship it back to Ivan.

Day: April 22, 2008

Tyan Transport GX28 (B2882)

Failing PLED?