Google Mobile Crawl errors

It would be harsh to criticize what is an excellent free service from Google in their "webmaster tools" service, but I found this amusing so I thought I'd share.

My desktop just installed the new udev (requires 2.6.18+), because a 2.6.25 kernel was installed, even though this kernel isn't running and doesn't boot on this hardware.

I better fix the kernel (and initrd.img)megaraid drivers before the next power cut, and get it right first time.... Although I'm kind of pissed that the kernel maintainers haven't applied said patch to the megaraid driver by now.

Leaving non-functioning stock kernels around is not a good idea in Debian.

Oh well this box is overdue for a reinstall.

We used a Squid reverse proxy hosted in a good data centre many moons ago.

We need a similar setup now whilst we sort out relocating current servers, and such like, to off load serving of static pages and other static content for some websites.

I (well Google) failed to find a quick and dirty reverse proxy server network being offered.

Everyone offering CDN wants me to ring them up (which I'm guessing we'll have to pay the sales staff salary as well).

We don't need the earth here, just a few megabits per second, ideally caches in both London and the US, London would do at a push, but unless someone knows of a sensibly priced third party solution, we'll get a virtual server or two from some good webhosts, and run Squid as a reverse proxy for a few months.

So before I start ringing up sales folk - who should be top of my list?

I'm figuring anything that requires more than "domain name" "IP address of current server", some sort of DNS delegation from us for the DNS records in question, and a credit card number, is too complex and we won't want to play. Anyone who hasn't automated the sale of the above is probably too expensive....

Warning: Cannot send session cookie - headers already sent by

A common error when using "session_start" in PHP is the "headers already sent". As the manual says some content got emitted to the browser already.

Usually the advice is put "<?php session_start(); ?>" as the first line in your PHP file (assuming it isn't included). This usually sorts the issue, but today it didn't work as expected!

Turns out the problem was I was using the webmaster's Microsoft Windows box with Dreamweaver and had switched it to use UTF-8 as the default character encoding, which was then quietly prefixing the files with the UTF-8 Byte Order Marker <U+FEFF>. All very obvious when I ssh'ed into the Debian box and did a "cat" on the file. Apparently some versions of Dreamweaver allow you to disable the BOM placement, I had to switch the page's encoding to "Latin-1".

Grr - guess I should stick to tools I know

The Quality and Outcomes Framework in the UK is designed to measure the performance of General Practice (doctor's surgeries).

I was somewhat skeptical since the points awarded for hypothyroidism are on the basis of knowing which of your patients are hypothyroid. Then additional points are awarded based on the proportion of those patients who have had a thyroid function test in the last 15 months. In my view, this is pretty much the minimum standard of care one might expect for someone with hypothyroidism, hence my skepticism.

However it is interesting to note that the prevalence of diabetes and hypothyroidism measured by the QOF has increased by about 10% each. The change is most likely mainly an artifact of the process, reflecting that doctor can now correctly report the existence of 10% more such patients. But it is a significant increase in number of patients known to be receiving at least the "minimum standard of care".

Whether this reflects a genuine improvement in the care received, or merely improvements in the ability of GP surgeries to report the care they are providing is another question. I trust it doesn't reflect a 10% increase in endocrine disorders.

Prevalence figures are based on number of patients registered with a GP, and may thus overstate the actual prevalence of a disease (since, we hope, people not registered with a GP are in generally good health).

http://www.ic.nhs.uk/cmsincludes/_process_document.asp?sPublicationID=1174988969611&sDocID=3163

My desktop box that is using testing (Lenny) was behaving strange with highlight and drag and drop not working quite right. Seeming to send more events than expected in KDE, getting in a mess every now and then.

After some reading around similar issues with touchpads (nope 3 button mouse).

I braved...

sudo dpkg-reconfigure -phigh xserver-xorg

Which backed up my /etc/X11/xorg.conf file and then ripped all the details out. On restarting kdm everything seems sensible so far.

I assume I'd edited xorg.conf causing dpkg not to update it when the latest xserver-xorg package was installed. The new xorg configuration trick is going to be to remember to make sure the configuration file is suitably devoid of meaningful content.

The more things change.....

Everyone is having fun working out the gender of their browsing habits.

http://www.mikeonads.com/2008/07/13/using-your-browser-url-history-estimate-gender/

The geek in me (after allowing it using NoScript) looked at the source, to see how it gathered my browsing history.

It opens an iframe with a list of URLs and checks the state of the URL to see if you have visited it. Immediately leaking information from your browser history to any site you allow to run JavaScript.

I knew that the same origin implementation in common browsers was poor, but I didn't realize how poor. The World Wide Web needs a reimplementation, I would say redesign but many of these issues were anticipated by the designers, it was just the implementors who cut corners.

Further reading;

http://crypto.stanford.edu/sameorigin/sameorigin.pdf

NoScript mentioned again -- oh dear I'm turning paranoid.

Whilst I've been enjoying watching my son grow amazingly fast this summer, others around me have been having an aestivus horribilis.

As such I demand everyone should blog only good news for the rest of the month.

I'm wondering did the kernel folk ever go back and review the split of the Megaraid driver into two? I understand why these kinds of changes sometimes need to happen, but seems to me that it caused a lot of practical issues.

I'm still left with machines where I (allegedly) have to manually patch the kernel sources to get the right Megaraid driver loaded (presumably because no one had a complete list of hardware IDs), and today I was upgrading an ancient system (to make a test server) and I needed to manually recreate the initrd image with the legacy Megaraid driver in (because it was omitted by default, and the root file system was thus not found). This kind of thing scares off administrators, and Google is little help unless you understand the topic as it mostly lists thousands of others having similar issues.

Meanwhile my bitter experiences of Adaptec, and LSI PC Hardware RAID solutions leads me to be jaundiced about hardware RAID on Linux. Sometimes a good UPS software RAID and keeping stupid people away from your hardware is a better solution. Waiting for disks to spin around is a stupid way of committing data to permanent storage, and the sooner something better becomes truly mainstream the better. Flash storage is pretty much there, but I think administrators need to be clear how much they need this. Our servers are almost universally limited by committing writes (except for the big hardware RAID one, where I can't diagnose what is making it slow because the kernel says there is no I/O at all, but I suspect it is Postgres writing too much too often (because I know how the software is written)).

Wise word follow...
http://linux.yyz.us/why-software-raid.html

I think NoScript is an excellent Firefox plugin, that makes surfing both faster, and safer. Today I hit my first problem in many months of usage. A website using Google-Analytics "__utmlinker" for navigation.

"__utmlinker" is a method of passing cookie details for Google Analytics old Urchin tracking system from one domain to another. Since I don't trust Google-Analytics (in NoScript at least!) the function isn't available, and the link just didn't work for me. End of sale? In my case I cut and pasted the URL out of the JavaScript and carried on.

The "free" Google-Analytics seems to have dampened down the web analytics market somewhat. Is cross subsidizing your Analytics offering with money from your online advertising, and other services, evil?

In my to-do list is "check out nuconomy.com" who include a server side API in their analytics product. It isn't exactly a complicated API, but it makes so much sense for sites like 350.com at work, where navigation doesn't tie nicely with purchasing, and we want many ways of getting to the same Perl purchasing code, just to put the conversion tracking code on the server, rather than worry too much about how they got there (surely that is what the analytics should tell us - without doing all the hard work ourselves).

It strikes me that Google (Urchin) shouldn't have needed or encouraged folks to put in JavaScript linking, since if they have JavaScript running in source and destination, they can fairly easily figure out if it is the same browser instance. Or they could have done something that degrades more gracefully, but then perhaps they assumed folks would trust them!

Either way, if you are starting to use 3rd party JavaScript in place of HTML anchor tags, perhaps it is time to find a simpler way.