I recently got a new MacBook Pro and one interesting thing i noticed was light “click” (a clicking noise) from it whenever it was idle for a few seconds. I pay attention to such things since I heard about problems with power management settings under Ubuntu, which could quickly wear down a hard drive. I experienced this myself, where one of my old hard drives started to sound like a frog :-/. So I installed smartmontools (either use MacPorts or fink) and checked:

As you can see I have a Hitachi 500GB 7200rpm drive. The puzzling fact here is the Load_Cycle_Count. You can see a value of 36,492 load cycle counts in 351 hours the HD was powered on, so approx. 100 per hour.
Put easily the load cycle count is how often your HD decided to park its heads. Depending on the manufacturer and HD model this can mean several things. In my case it means the number of times the HD’s heads are moved to a ramp next to the platters. The advantage of this is that being in this “parked” position the drive can shut down some energy consuming parts and it is much harder to damage the drive when the heads are parked (nothing there for a Head crash).

Ramp Load/Unload Dynamics (c) Hitachi

To observe the development of your Load_Cycle_Count you can use the terminal with this small one-liner:

The script will log the load cycle count to your terminal and a file called hddLoadCounts.log in the current directory every minute.

You might notice that when doing nothing but browsing this count increases by 2-8 every minute. Playing music with iTunes seems to stop this, as the HD keeps busy reading your music. Doing the maths you’ll find that it’s not unlikely that your drive will have over 300,000 load cycle counts withing the first half year (lucky music listeners, yours will last much longer ).

As I had a bad feeling about this, i went on to have a look into Hitachi’s technical specs for my HD. Here you can find that my HD is designed for up to 600,000 load cycles (page 2), meaning approx. 6000 hours at the 100 cycles per hour rate. In the specs on page 135 you can find that if Advanced Power Management is enabled, the deepest reachable power saving is depending on the Power Management level. In general the Advanced Power Management Level is between 1 (power saving) and 254 (performance). If the Level is 0 or 255 no power saving is done, if the level is 1-127 it’s “Standby”, if the level is 128-191 it’s “Low Power Idle” and if it’s 192-254 it’s “Active Idle”.

As we’ll find out in a second, the default value (which Mac OS X sometimes seems to reset) seems to be 128, so “Low Power Idle” mode. The three power saving levels are explained in Section 12.6 “Advanced Power Management (Adaptive Battery Life Extender 3) Feature” of the specs. In short: “Active Idle” mode cuts down power consumption by 45-55%, the heads are parked near the mid-diameter of the disk, recovering takes about 20ms. In “Low Power Idle” mode power is cut down by 60-65%, the heads are unloaded to the ramp (this is the “parked” counted by Load_Cycle_Count), recovering takes 300ms. Transition into these modes is magically done internally by the HD (it observes what’s going on and decides what to do next), taking into account the Advanced Power Management Level. (“Standby” mode isn’t mentioned here, but it sure unloads the heads to the ramp, as it spins down the HD… recovery will take long, but unimportant, as we’re having a problem with “Low Power Idle” mode.)

So how do we find out which Advanced Power Management (APM) Level our HDD uses?
This doesn’t seem to be very easy in Mac OS X as there’s nothing like the hdparm on Linux.
There is the hdapm tool, but it can’t read the value, you can just set it. We’ll learn why this tool is necessary in a moment, but first let’s find out what the current value is.
The easiest way to accomplish this was to throw in a Linux Boot CD (Knoppix, Ubuntu, whatever you like), reboot, boot from CD (hold down the “c”-key), then fire up some terminal, become root (sudo -i and check the current APM value:

For me it was 128.
You can check the immediate effect from within the Live CD: you can use smartctl -a /dev/sda. As before this kept increasing.

As I always handle my laptop with care and can live with 10 % more power consumption of my HD, I decided to change the default. WARNING: This might not be suitable for you, it’s your decision.

To stop this rapid growth of the load cycles, I first tried to set the value to 191, but i could still observe a rapid increase.
After setting the value to 192, it immediately stopped:

Afterwards i rebooted, the Load_Cycle_Count increased by 1 over the reboot and no more after a couple of hours runtime (without iTunes keeping my HD busy, draining my battery). To my surprise the next day my logs showed that the load cycle count was increasing rapidly again, i rebooted back into linux and found the value was reset to 128. Weird. I reset it, rebooted, the count didn’t increase anymore, but at some point i again found it increasing rapidly. Based on this I assume Mac OS X or something else (like Windows run via bootcamp) sometimes resets that value to 128. My first guess was that maybe it is reset after resuming from sleep, but I couldn’t reproduce it by this. If someone finds out let us know in the comments.

To overcome this problem it seems sufficient to have a tool which explicitly resets the APM level once during system startup to something meaningful. That’s where we remember the hdapm tool: You can download it from the given page and install it as described in the user guide.
Afterwards edit /Library/LaunchDaemons/hdapm.plist to set the correct APM value. As a reference my file looks like this:

Notice that for other drives, especially other manufacturers the 192 might not be the right value. If you found the correct values for other drives, be welcome to share them (preferably with links to tech specs) in the comments.

Now, after four months since discovering the problem went by, my load cycle count only increased by about 500 (which is about the number of times i sent the mac to standby). Isn’t that a figure compared to the 36500 within the first month?

Edit (Sep. 22, 2011): Revised my guess about sleep causing the reset. Thx to Sam.

Digging into I found out the framework is used for live mapping (collection, aggregation and visualization) of disaster and event related messages sent via all different kinds of transports (e.g., twitter, facebook, email, sms…). One example is the 2010 Haiti earthquake. Where it helped to coordinate all the s&r teams.

As I find it quite fascinating how much people who sit at home in their living rooms might be able to help others in a disaster region, I’d like to suggest this talk:

The project can be found on github: http://github.com/joernhees/thesis-skeleton.
I’ll happily include / pull changes.

Quick instructions to get started with your thesis:

1. Make sure you have git, otherwise install it (e.g., on ubuntu: sudo aptitude install git-core)
2. Run this:
   git clone git://github.com/joernhees/thesis-skeleton.git myMasterThesis

   It will create a directory called myMasterThesis in the current directory which actually is a git repository and includes a thesis directory.

3. Enter it and have a look at thesis.pdf
4. Insert your name, title, supervisors, etc. in thesis.tex.
5. Get familiar with git, this is a good start.

That’s it.

This time a very interesting analysis of what happens when he posts on his blog and twitters about it.

Most interesting: the number of bots that access his page just seconds after he published it, where did the payload of human readers come from and what did it change in numbers of subscribers.

The BetterRelations Game in action (click to play)

Enjoy

(also see the first post)

BetterRelations (alpha)

The BetterRelations Game in action (click to play)

The game collects pairwise user preferences, which are then used to rate Linked Data triples by “Importance”. Would be cool if you find time to play the game maybe in the lunch break and help me collecting the data for my thesis.

Feedback and bug reports are heartily welcome. If you know other interested players feel free to forward the link or this post, the more people, the better

More to come, keep posted.

The last attempt actually wasn’t that wrong, as it only lacked the errors="ignore" flag:

One might think we’re done now, but this depends on your Unicode Normalization Form: Unicode allows Combined Characters, for example the precomposed u"ü" could be represented by the decomposed sequence u"u" and u"¨" (see Unicode Normalization).
In my case I know that my unicode strings are in Unicode Normalization Form C (NFC) (at least the RDF Literal Specs say so. This means that if there is a precomposed char it will be used. Nevertheless Unicode potentially allows for Combined characters which do not have a precomposed canonical equivalent. In this case not even normalizing would help, multiple unicode chars would remain, leading to multiple multi-byte UTF-8 chars. In this case I’m unsure what’s the universal solution… for such a u”ü” is it better to have a u”u” or nothing in case of a split? You have to decide.
I decided for having an “u” in the hopefully very rare case this occurs.
So use the following with care:

Unicode and UTF-8 is nice, but if you don’t pay attention it will cause your code to contain a lot of sleeping bugs. And yes, probably I’d care less if there was no “ö” in my name

PS: Günther, this is SFW. :p