Jonathan's blog

Fedora 13 (“Goddard”) was released today.

I wouldn’t normally go upgrading my OS to the latest on the day of release, but frankly anyone who runs Fedora is an early adopter by definition.

I started by upgrading two unimportant Fedora 12 virtual machines at work using preupgrade. One went smoothly but the other failed because the /boot partition was too full. I cleared out all old kernels and tried again, with success. Each upgrade took less than an hour, I think, but I wasn’t really paying attention.

After brief testing to make sure all the important stuff had upgraded properly, I upgraded my work desktop PC, my home PC and my laptop too. They were all on Fedora 12 and the upgrades went without a hitch. I’m very impressed.

Massive thanks and kudos are due to the Fedora team for working so hard to get this release out and for providing such an easy upgrade path. I look forward to getting stuck into the new features of this release in time.

My next task is to upgrade my home server, which is currently running Fedora 11. Updates for Fedora (N-2) are only available for one month after the release of Fedora N, so time is now of the essence if I wish to keep my server secure. Unfortunately the reason I’m still on 11 is because the upgrade to 12 failed and I wasn’t able to get it working. I will probably take this opportunity to do a complete wipe and reinstall (scary!). Then I can also migrate from i386 architecture to x86_64.

Watch this space!

For those who have been following the issues around Samba suddenly breaking upon upgrade, I’ve now got to the bottom of it.

I’ve updated the original post with details, so if you wish to leave comments, please leave them on that post.

Some time ago I bought a cheap USB thermometer called TEMPer. I was disappointed to find that it didn’t work on Fedora. It would only work on Windows using a poor piece of proprietary software.

I eventually found the blog of Tollef Fog Heen, who had managed to get his TEMPer to work. Unfortunately his solution involved patching and compiling a kernel.

However since then, it seems his patch has been integrated into the stock Fedora kernel and it is now possible to read the temperature from it.

The TEMPer device appears to be a USB-serial adapter, with a serial I2C device at the end of it. It’s not straightforward to extract the temperature from it, but Tollef Fog Heen has written a simple C program to return the temperature.

His program polls the TEMPer every second and prints the temperature to the command line. It doesn’t stop until you kill the program. I made a couple of tweaks to the code so it prints the temperature once, formatted as a raw number with no extra text, and then quits. You can find my modified source here.

Disclaimer: I don’t know C. I haven’t changed any of the logic of the code, only the way it prints the output. If the code is buggy, it wasn’t me!

Now I have an executable that returns the temperature from the TEMPer, I can think about building some application that could use this. How about a Nagios plugin?

Yesterday at work I had the need to create a federated table in MySQL. I read about the federated engine and thought I had it sussed. I noted:

Beginning with MySQL 5.1.26, the FEDERATED storage engine is not enabled by default in the running server; to enable FEDERATED, you must start the MySQL server binary using the --federated option.

Turns out it’s also possible simply to add the line federated in the [mysqld] section of /etc/my.cnf

The version of MySQL currently installed on my CentOS box was an older one (5.0.45) but I added this line anyway. The server refused to start. It quickly became clear that the MySQL binary packaged with CentOS was not compiled with the federated engine.

Fedora is currently packaging MySQL 5.1.37 but it seems that this too is lacking the federated engine. That’s annoying – I had wanted to install a version of MySQL from some yum repo or other, so I don’t have to keep upgrading the package every time a new version is released.

Perhaps the lack of federated support is a Red Hat (and derivatives) issue. I downloaded the rpm from MySQL directly, and installed it. Guess what – no federated engine compiled in.

So I downloaded the source tarball. I explicitly configured it with the federated engine, like so:

./configure --with-plugins=federated

And then I built and installed it. Nothing worked properly out of the box, and I was annoyed to find that the make install command doesn’t do half of the things I would normally expect it to do. I found this information and followed the steps to get it working. I had to steal and tweak the /etc/init.d/mysqld script from a different box which was running the bog-standard CentOS package.

Woohoo! The federated engine was finally available.

mysql> show engines;
+------------+---------+----------------------------------------------------------------+--------------+------+------------+
| Engine     | Support | Comment                                                        | Transactions | XA   | Savepoints |
+------------+---------+----------------------------------------------------------------+--------------+------+------------+
| InnoDB     | YES     | Supports transactions, row-level locking, and foreign keys     | YES          | YES  | YES        |
| MRG_MYISAM | YES     | Collection of identical MyISAM tables                          | NO           | NO   | NO         |
| BLACKHOLE  | YES     | /dev/null storage engine (anything you write to it disappears) | NO           | NO   | NO         |
| CSV        | YES     | CSV storage engine                                             | NO           | NO   | NO         |
| MEMORY     | YES     | Hash based, stored in memory, useful for temporary tables      | NO           | NO   | NO         |
| FEDERATED  | YES     | Federated MySQL storage engine                                 | NO           | NO   | NO         |
| ARCHIVE    | YES     | Archive storage engine                                         | NO           | NO   | NO         |
| MyISAM     | DEFAULT | Default engine as of MySQL 3.23 with great performance         | NO           | NO   | NO         |
+------------+---------+----------------------------------------------------------------+--------------+------+------------+
8 rows in set (0.00 sec)

But I can’t understand why none of the binary builds of MySQL include it. Fair enough that isn’t enabled by default in the running server – it’s no problem to add a line to my.cnf on a standard CentOS box. But it is a nuisance to have to build from source. It doesn’t break anything to have it enabled in a build, even if unused.

Of course CentOS won’t change the way they build their packages until Red Hat does. So I’m doing what I can, and I have filed a feature request with Fedora in the hope that in the next major release, there will be a version of MySQL built with the federated engine.

I decided that I need to sort out the way I do my personal calendaring.

Currently I only use my phone’s built-in calendar. I nearly always have my phone with me, but it’s a bit of a pain to enter stuff on when I’m sat at a computer anyway, and carrying all that information solely on my phone presents a huge risk of loss, theft or breakage.

I need some kind of centralised store of information that is able to sync with all the devices and programs I want to use, namely:

• Some sort of cross-platform calendar client – mainly for use on Linux but also nice to be able to use similar software if I’m on Windows or OS X.
• Sony-Ericsson P1i (Symbian) built-in calendar
• iPhone, for when I get one
• Web interface, for those times when I’m borrowing a computer and can’t install a client.

Google Calendar seems to be a good choice. It’s flexible and can sync with lots of things.

Linux

So I installed Lightning on all my Fedora and Ubuntu machines. It’s a calendar extension for Thunderbird, which I already use. To install it yourself:

On Fedora:
yum install thunderbird-lightning
On Ubuntu:
apt-get thunderbird-lightning

It’s easy to set up, too. Suppose your Google account is joebloggs@gmail.com, then you would…

• Add a new calendar to Lightning by right-clicking in the Calendar area
• Choose On the Network
• Select CalDAV
• Enter your location as https://www.google.com/calendar/dav/joebloggs@gmail.com/events
• Give the calendar a name

OS X and Windows

It’s a little more work to install Lightning on OS X. You have to download the add-on from Mozilla, and install it in Thunderbird. Same story for Windows.

It’s quite straightforward and there are instructions on the website.

When you’re done, follow the same instructions as for Linux to subscribe to your Google calendar in Lightning.

Sony Ericsson UIQ

Setting up Google Calendar on my Sony Ericsson P1i was a bit of a pain, too. The P1i can’t interact with Google natively, I had to set up an account with Goosync to enable this. Goosync talks to Google, and your phone talks to Goosync using SyncML.

But once you have a Goosync account, you can synchronise a lot of handsets with Google calendar.

So first, you will need to set up an account with Goosync. It’s free and very easy. Goosync will ask you to tie your Goosync account to your Google account.

There’s also an option to have the settings for your phone sent automatically to your handset. However this didn’t work for me so I had to enter the settings manually.

Assuming the sync task on your phone has been set up properly, do a  test run to make sure it all works.

• If possible, connect to a wireless network first. If not, 3G will have to do.
• Go to the Main Menu
• Go to Tools
• Go to Remote Sync
• Find the profile that syncs with Goosync
• Find the sync task called Calendar. Make sure it is ticked, and then tap Sync to start off the first synchronisation.

If that worked, you can now run the sync task whenever you like from within the calendar itself.

• Open your phone calendar
• Tap More
• Tap Calendar manager
• Tap Synchronise

That’s all there is to it! Unfortunately there’s no way of making your calendar synchronise automatically at set intervals, but that’s probably a good thing, because you can’t get stung for 3G charges!

iPhone and iPod touch

Coming soon…

If you have Fedora and an nVidia graphics card, chances are you’ll want to use kmod-nvidia as your graphics driver. It is closed-source, but produced by nVidia themselves and has several advantages over the default open-source drivers that are typically bundled with most distributions – for example, 3D hardware acceleration.

If you have already installed kmod-nvidia – read on, and find out why you should upgrade to akmod-nvidia.

So what’s wrong with kmod-nvidia?

The way it works is simple. For each kernel version, there is a corresponding nVidia kernel module. Keeping the two in sync is a pain, so the packagers at RPMFusion have made a metapackage, simply called kmod-nvidia which tracks the right version of the module for your kernel, e.g. kmod-nvidia-2.6.29.5-191. It’s simple – you install just the metapackage and yum automatically installs the right version of the kernel module.

The problem arises when a new kernel is released, but the packagers of kmod-nvidia haven’t yet released the corresponding kernel module. Sometimes they do it in a few hours but often it takes longer – maybe a day or two. For all the time that the corresponding kernel module doesn’t exist, you cannot update your kernel (and if you are using PackageKit to update your system, you cannot easily update anything!)

What’s different about akmod-nvidia?

akmod-nvidia is different. Rather than downloading someone else’s kernel module when it’s available, akmod-nvidia compiles its own version of the module for whatever kernel you have.

So if you update your kernel, next time you boot into the new kernel, akmod will see that no module exists yet on your system for your kernel, and it will compile it automatically. This takes only one or two seconds – I haven’t noticed the delay on my system.

The advantage is that you don’t have to wait for anyone else to do anything when you update your kernel. It’s also extremely useful if you are running some sort of custom kernel, such as PlanetCCRMA‘s realtime audio kernel.

So how do I install akmod-nvidia?

If you haven’t already got the RPMFusion repository set up on your computer, you will need to do this. (The following code snippet is for Fedora. For CentOS, see the RPMFusion Configuration page.

rpm -Uvh
http://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-stable.noarch.rpm
http://download1.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-stable.noarch.rpm

If you already have kmod-nvidia, uninstall it.

yum remove kmod-nvidia

Then you can install akmod-nvidia. It will probably need to pull in a handful of dependencies.

yum install akmod-nvidia

Now if you reboot, akmod will automatically compile your kernel module. You’ll never have to wait for packagers again!