A while back I was asking about how to look at server load issues. I wound up using collectd, which was pretty useful.

I identified a handful of disk I/O spikes - unfortunately it’s hard to match these up against the subjective delays/lag problem. However, I looked into ways to improve I/O handling, and found information about caching and pdflush.

I’m currently experimenting with upping /proc/sys/vm/swappiness (to 90), to encourage more swapping to disk - on the grounds that I do have an active server, and I’d rather use the disk cache more. I think. I’m a little unsure of my logic here, but so far this seems to have improved the situation a bit, so maybe I’m on the right lines.

In When Do You Trade in Your Gibbon for a Heron?, I mentioned that I’m considering upgrading my System76 laptop from Gutsy Gibbon to Hardy Heron. A commenter named Scummy suggested that a similarly configured Dell system is cheaper:

Dude - you just paid a $350 ‘Linux Tax’ by NOT going mainstream in your hardware…

Maybe so, but I think not. It depends on what you value.

Fedora 9 is now available - http://get.fedoraproject.org/ As someone thats been working with it for months already I’m quite pleased with this release.

My main computer these days is a laptop from System76. I was very happy to support a vendor willing to ship and support a laptop without the Microsoft or Apple tax, and very pleased to have hardware supported by a Linux distribution.

That was a couple of Ubuntu releases ago. Now that Hardy Heron is out, I feel the urge to run through the upgrade. Improved battery life is one big draw, but getting updates for all sorts of software I rely on is nice. (New Wine, new Valgrind, new Firefox, new GCC….)

Upgrades are never 100% perfect, though. There’s always a risk of mishap that’ll take me a few days to straighten out. I don’t have any talks to give or books due in the next several weeks, so the timing seems good, but perhaps I should wait a few days more for the early adopters to expose any bugs that I’m happy not to fix?

I know plenty of people like Jim Zemlin who were very happy to run pre-release versions of Hardy Heron, but he had Greg K-H to help him out.

How do you decide to upgrade?

10 days ago the Linux Loop blog had a post titled “Linux Eee PC Far Faster Than Windows Version”. I’m sure many Linux users nodded and had assumed as much. The author compared the times of three tasks: boot up, loading Firefox, and shutting down. That’s hardly a comprehensive set of tests. Some people commented to dismiss these metrics as “meaningless”. They aren’t meaningless but they certainly aren’t the whole story. Others pointed out that IE on Windows was faster than Firefox on Linux and that MS Works was faster than OpenOffice. Some then responded that Works isn’t the equivalent of OO and that MS Office would be a better comparison. It all got a little shrill with those who believe that Linux is faster than Windows and those who say it isn’t so talking past each other and resolving nothing.

I’m going to try and sift through the morass and say what I think the numbers really mean and what they don’t mean. Those with an agenda, either agenda, will, I’m sure, attack what I have to say. I think anyone who really tries to look at things objectively probably won’t. I’m just not sure that very many people are truly objective.

In the interest of fairness let me disclose where I am coming from: Yes, I tend to have a pro-Linux bias. I also have a bias against hype and B.S. When it comes to my professional life I’m an IT mercenary. If someone wants me to support Windows systems along with Linux or UNIX systems I will gladly take their money and do the work. I also won’t evangelize on behalf of Linux. Why not? In business everything comes down to a cost vs. benefits analysis and there are situations in the real world where a change of OS is far too costly to justify any perceived benefits. There are many situations where Windows or commercial, proprietary UNIX really and truly is the best fit. Back to those pesky metrics…

There are some “interesting” issues being raised due to an incompatibility between e2fsprogs and GRUB legacy that results in non-booting systems. Newer versions of e2fsprogs default to creating new Ext3 filesystems with 256-byte sized inodes, instead of the old default of 128. GRUB legacy has absolutely no clue what to do with 256-byte inodes, so it barfs up the “Error 2: unknown file or directory type” message and sits back down. GRUB legacy will not be patched to support 256-byte inodes. Yay! But this is a transitional problem; read all about it here:

GRUB vs. the Inodes: Who Needs a Bootable System, Anyway?