Spent a fair chunk of today sorting out Phil’s hard disk. Mac OS X had impolitely destroyed his partition table for him. I used TestDisk to recover it. There were a couple of issues, which meant that it took a little longer than it was “supposed” to.
It located a couple of adjacent ext3 partitions (Phil’s Fedora and Ubuntu roots) and generated some overlapping numbers for them. I had to manually edit the partition table using sfdisk to get it right. I particularly enjoyed the rollback-like features that sfdisk has (using -O and -I). I had to tell sfdisk to use sector units because it defaults to cylinders.
Then Phil gave me food. Excellent. Technical support for food.
Got a couple of management textbooks out from the library earlier. They’re really boring. Seem to be stating the obvious. Joy.
Overview of last week of stuff:
- I started the week with little understanding of the Linux ARM boot process, Steve’s NSLU2 and the need to get a working Linux install on it. Tom and I spent a considerable amount of time working on getting it to work. Almost all of the problem was because for some (yet to be fully determined) reason, the kernel is passed an incorrect machine code by redboot. Whether this is because someone’s flashed the slug in the past, I don’t know. I ended the week with an understanding of the ARM MMU, linux ARM boot process and a working booting slug with functioning network adapter :-)
- I created my first RPM. :-D I spent christmas eve reading various documentation on the subject, and so have packaged my yum plugin. I’ve stuck the spec file, srpm and rpm up for public access.
Over the past few days I’ve been working on a yum plugin that will pick-up local yum repository mirrors using avahi. It wouldn’t have taken that long to sort out if I hadn’t failed to read the bit of the dbus tutorial I was reading that said I’d need “import dbus.glib” to make callbacks work.
I spent many hours trying to work out what was going on. Steve found the problem with me not importing dbus.glib. Thanks Mr Steve.
I’ve put the code up here. It’s a git repository. Grab it using:
git clone http://www.ecs.soton.ac.uk/~rds204/yum-avahi/.git
There are two python files: repos_announce.py and repos_listen.py
repos_announce.py: This is to be run on the server. It uses avahi to tell the rest of the network that a yum repository is available locally. It points to the http server that it runs which servers the “conf” file, which contains the configuration for the various yum repositories.
repos_listen.py: This is the yum plugin. Stick it in the yum plugins directory (defaults to /usr/lib/yum-plugins). The yum plugin also requires a configuration file called “repos_listen.conf” which sits in the yum plugin config directory (/etc/yum/pluginconf.d), and this needs to contain:
enabled = 1
Currently it will only attempt to grab repositories that you already have configured.
I reckon it’s a relatively safe thing to have enabled even in potentially unfriendly networks, as long as you have gpgkey checking enabled.
Quite a difficult one this. I have been a user of Fedora for a couple of years now, and it very much feels like home. Times are changing, and I can feel the pull of Ubuntu.
One major thing that sways me is Tapioca. I’ve tried the rpms from the repository that the Tapioca site links to. They install fine. They just don’t work. I have not succeeded in having a conversation using Tapioca yet. I’ve visited the IRC channel and spoken to people about it, and they all appear to be using Ubuntu. Ubuntu with functioning Tapioca :-(
However, there are some things that push me towards Fedora. Fedora appears to have decent test procedures in place for packages and stuff, which, despite Fedora being ‘bleeding edge’, means that things tend not to break. In my experience of using Fedora, it’s only broken once, and that wasn’t a fatal break: booting didn’t work under the new kernel, but it did under the older one (fix: use older kernel!).
Then there’s this whole other debate about yum or apt. As far as I can tell there’s no real feature difference between rpm and deb (except for deb’s “other packages you may be interested in” feature, which to me seems a little superfluous). I haven’t had any trouble with yum on my machine. Howard managed to corrupt part of rpm by switching his machine’s power off during an update, but still that wasn’t anything to do with yum.
Justyn, a housemate, seems to complain about yum continuously and I just don’t know why. I run local mirrors of the fedora repositories to save bandwidth (there are now 5 people in this house who use fedora), and have absolutely no trouble in using or maintaining them. Justyn complains regularly that yum seems to fail and die. I’m still waiting for him to show me such a case.
My one complaint about yum is that it’s slooow. The situation is apparently a lot better in FC6 because it uses yum 3.0, which has moved the xml parsing into a C library. apt seems to be uber-fast, which is good.
I’ve read and heard a lot of people say “Ubuntu just works”. This is great. However, I read that apparently this has a lot to do with the Ubuntu developers hoarding patches and not sending them upstream.
It looks like I’ll stick with Fedora for a while yet. I feel the need to experience FC6.
|<< Newer Posts|
Site by Rob Gilton. © 2008 - 2019