I thought apt-cacher was an improvement on apt-proxy. Boy was I wrong. It’s been giving me a lot of headaches. Whereas apt-proxy needed the odd restart, apt-cacher needs files to be deleted from it’s cache every now and then…
What a pain.
Now I’ve switched back to apt-proxy, and with a bit of squid tweaking (so that local network traffic isn’t cached, or passed to the SAIX proxy) everything is working well again.
I’ve touched up the CLUG wiki GPRS/3G page today. It’s now a complete dump of my knowledge on the subject - a page I wish had existed when I started mucking with mobile data.
Now I’m just waiting for the 3G card that I bought on eBay to turn up, so I can get a slightly more usable connection than asthmatic GPRS when I’m on the run.
Somehow, I’ve managed to pick up flu during the heat-wave, and been out of action for the last week. While everyone else has been dying of heatstroke, I’ve been wrapped up in arctic gear, complaining of droughts, and generally sleeping 24hrs a day. :-) I’m slowly getting back on my feet now, thank god.
Now that I’m back online, I’m seeing lots of e-mails from
sensords on servers, complaining of being too hot. I’m not the only one with a temperature :-)
I decided I wanted a 64bit kernel for my UK2 server (seeing as it’s an Athlon X2, and I wanted an ubuntu amd64 chroot to work on it).
After a little playing around, this is how you do it:
make menuconfig ARCH=x86_64
linux-amd64-ld, etc. wrappers. Some need
-m64options. Others autodetect what’s required. (I made a tarball of my wrappers, but they’ve been lost.)
CONCURRENCY_LEVEL=4 make-kpkg --append-to-version '-londonpride' --revision '1.00' --rootcmd fakeroot --arch amd64 kernel_image
Unfortunaly, in the week after this, my machine died of disk corruption. This is the 2nd time it’s happened. I hope UK2 don’t use dodgy hardware…
Now that I got my laptop back, I’ve been trying to get it to work with Ubuntu Edgy. I tried a lot. I spent hours pouring over kernel command line options, to stop the random hangs I was experiencing due to bad interrupt configuring.
Eventually I gave up and decided a new kernel is the only solution.
Ubuntu has a great custom kernel guide, and a git guide. The build took 2 hrs. and Git downloaded well over 100MB. Then I tried to build a restricted modules, and after fighting for a few days, gave up in disgust.
My new kernel suffered from the dreaded ati-ixp double speed clock. I tried every option, and in the end,
no_timer_check seems to be the one to use.
So I put this in my
And after using that a bit, I upgraded to fiesty anyway, and bricked my system, thanks to a lovely initramfs-tools bug. I now carry around a monolithic kernel (like all my debian boxen use). You just can’t trust initrds…. :-)