According to Debian bug #399113 and linked discussion it is impossible to run a stable system on Xen without enabling PAE. It seems that no-one is considering the fact that a hypervisor that runs on both 32bit and 64bit architectures should be able to support 32bit systems with 4G of RAM (IE not using the PAE feature).

But instead to work around this bug the Debian developers have decided to just enable PAE. This is annoying for me as I have to either buy a new laptop or reduce my use of Xen.

I wonder what will happen if a Xen bug is discovered that only happens on PAE systems? Would that make Xen only an AMD64 thing?

I have not been blogging for some time now, due to several things:

• My server’s hda is dying: hda: dma_intr: status=0×51 DriveReady SeekComplete Error . Yeah, I hate hardware.
• I spend more time IRL than on IRC. Some time ago I left some (most of the) channels I used to hang in, and decided to spend more time with beers arounds than in front of the computer.
• All kinds of things happening in my life have led to a situation that I want to dedicate more time to play music and DJ (i already got word of a real gig some time soon), meet friends and go out clubbing than writing blong entries and doing Debian packages (still, I manage to update most of my beloved ones, except abcde, that needs a last upload but I dont know if it will make it for etch, eventhough the package has been ready for months).

On unrelated news, if you can stand a bit of disgusting scenes and a bit of perversions of the original book, you HAVE to go see Perfume. WOW.

DebConf8 will be in Argentina... :D

Improve your tango.... and be our guest :)

The GNOME Users and Developers European Conference (GUADEC) invite you to participate in the 8th annual conference on the 15-21st July 2007 in Birmingham, England. The deadline for proposals is Monday 12th March. Successful candidates will be selected and notified by the GUADEC organising committee. Unsuccessful candidates will still have an opportunity for their session to be scheduled during the Approach Weekend or After Hours. For more information, and to submit a proposal, please go to http://www.guadec.org/callforpapers.

I just decided I will leave most IRC channels now, for hopefully not more than a week. I'm busy with $work$, or rather I need to be more busy with it, and IRC is too much distracting me, resulting in less work getting done and being to hectic and harsh on IRC as well.

You can query me on IRC, but usually I prefer mails. See you in a while...

* Things are going slow & smooth here. Except, I was too much shocked by death of my grandmother at Palanpur. I have already written about this at my Gujarati blog. * LDTP, Festival-* already entered in unstable and I need to fix why it is throwing ‘dh_pysupport -a commad not found error’ in all architechtures. * No connectivity. Hope that this will resolve soon

Our despicable prime minister said "I think that would just encourage those who wanted completely to destabilise and destroy Iraq, and create chaos and victory for the terrorists to hang on and hope for (an) Obama victory".

That statement is wrong in many ways, firstly the US (with Australian help) has destabilised and destroyed Iraq already. Claiming that removing US troops at this stage will do any further harm is unproven (and except for the Kurdish area extremely unlikely). Trying to link a presidential candidate with a terrorist organisation is the worst type of dirty politics.

Texas Republican senator John Cornyn requested that the Australian government stay out of US politics. It sounds like a reasonable request. But I wonder whether John Cornyn got involved when the US ambassador spoke out against the leader of the Labor party at the last Australian Federal election...

Senator Obama suggested that if John Howard wants to criticise his policies in Iraq then he should first send another 20,000 Australian troops (there are currently 140,000 US troops compared to 1,400 Australians). According to the CIA World Fact Book Australia has a population of 20.3M while the US has a population of 298.4M. If the ratio of serving soldiers was to match the populations then Australia would have 9,500 troops in Iraq. If Australia was to be a full military partner of the US (but scaled down due to the smaller population) and therefore had troops in South Korea, Japan, and Germany then the number of Australian troops committed might be closer to the 20,000 number cited.

But this is no criticism of Senator Obama. For a number that was made up to demonstrate his point that Australia is not pulling it's weight in Iraq it's a quite reasonable estimate.

The US Democrat Senator Ron Wyden correctly described Howard's comment as "bizarre". This may become a catch-phrase for the Howard regime (as the Bush regime is described as a "miserable failure").

Johnny seems to think that it's in his best interests to send Australians to fight a war that has no benefit for Australia (or the US for that matter), but he doesn't have the guts to do it properly. Fortunately for him the Bush regime understands his position and allows him to send a token force to demonstrate support without the risk of any significant number of casualties or having to conscript soldiers.

This mutual meddling in elections is a demonstration of the way Bush and Howard conspire against the interests of the Australian and US citizens. Both countries need governments that look after the interests of their citizens at home and let the UN take a larger role in world issues.

News flash: gamers and Tron fans everywhere can rejoice; Armagetron Advanced will soon be up-to-date once more in Debian! The last time a new upstream version (0.2.7.0) was uploaded was almost two years ago now, and while the Debian package has languished in neglect, upstream has chugged on, fixing bugs and adding new awesomeness to the game. Just recently, the old maintainer orphaned the package, and since I use it anyway I’ve adopted it and given it some much-needed love. The new version is 0.2.8.2.1, what an, er, long version number. The new packages have been renamed from armagetron[-foo] to armagetronad[-foo] to reflect upstream’s name change. They include upstream improvements such as the addition of maps and improved graphics code, as well as packaging improvements like a new initscript for the server and rewritten, up-to-date manpages. While the new packages have been uploaded to experimental, they’re stuck in NEW due to the binary name changes. So, grab the x86 or x86-kfreebsd packages from here and test, test, test! (Yes, I run a kfreebsd machine. I wanted something eclectic for a non-essential, moderately old dual-core x86 box. Also, I’ll probably compile amd64 packages soon, but the amd64-machine I can use to build them on is currently in a half-installed state, which I need to fix before that. You can always build from source.) I’ve cleared with the security team that it won’t get in the way of anything and will probably be uploading to unstable as soon as there’s been some more testing to look for major problems. And since the newest version contains fixes for the security bugs affecting the old version, I’m going to snoop around and judge the [in]sanity level of porting back the security fixes for etch, since the new version will at most be unofficially backported to etch. Go, er, take a break in between fixing RC bugs. And test, test, test!

Frustration...

No Primary MX for a while

Since last thursday (2007-02-08) my main MX server was stopped to get some maintainance (IDE problems, could be the hard disk, IDE cables or something else, amazing ahm?), anyway, we moved the important stuff away from the line, did our homework testing the secondary MX and stopped the machine. Until now the DataCenter didn't told me when the machine should be back and I'm really hoping to get it back by tomorrow (2007-02-12).

In the meanwhile, the messages are safe, waiting in a temporary queue, we are not expecting to have a downtime of four long days. Anyway, I will need to catch up with my mail in the next days, I'm trying to keep up-to-date using the web archives and doing l10n stuff for Debian. :-(

Ma ana arrrrrrgentina will probably be annouced offically as the location for DebConf8!

Kudos to the Venezuelian team for their (almost equally) great bid and very fair behaviour in the decision meeting! In the end of the meeting when I saw .ar as the winner (not by much, but enough to be willing to go for it), I feared a painful end, but no, the opposite happened.

Thank you all. And now let's make DebConf7 a real fun and productive event

Interesting point from Network World entitled Wikipedia: On the skids?. Apparently Florence Devouard, chairperson of the Wikimedia Foundation board, said that the recent US$1M donations round will only cover a portion of the site's US$5M operating expenses. I find the numbers a little funny -- according to the Wikimedia Foundation finance report, operating expenses for 2006 were around US$800K. I realize that the site's traffic has nearly doubled since this time last year, and that they're hiring more full time staffers, but that still puts the project's expenses at under US$2M. I'm not sure I understand how the US$5M number was calculated. I think the Wikimedia Foundation could be a much bigger and globally more important organization. With advertising on Wikipedia, to the tune of an estimated US$100M, the Wikimedia Foundation would have enough money to support its Web site and tens of millions more to buy Open Content paperback books for poor children in developing countries. tags: wikipedia advertisement florence devouard wikimedia foundation

Curves are again with the "right" slope. Thanks to Neil Williams upload for dpkg-cross. Good news also for libapache-mod-backhand who has been uploaded today. We're now left with atokx, atokx2 and wmmoonclock. Today, I worked on remaining issues of various ways to "break" PO files in a few packages. This was affecting the import of material in i18n.debian.net which I intend to get as clean as possible. That will mean a small bunch of NMUs for these package as they currently require hacks in the sync scripts, which I don't like. Ah, and Rapha l Poir e got a very last medal in biathlon..:-)

In my previous post ( to this Debian section), I've mentioned we were preparing an event focused on direct contribution. Well, we've done it and it was very successful!. The event name we've choose was invol crate+, which means "involve yourself" and the plus sign meaning "more", since that was the main idea behind the event. We've worked for a couple of months, this includes live and online meetings, (even we all live in Lima sometimes it's quite difficult to match our spare time) and a lot of online work. We've just published the event report (currently in Spanish only, translation in progress). The main things I'd like to highlight is the fact that we got people doing real work and breaking the ice from "I want to do something, but I'm quite not confident to get on the teams" to people discussing work-related issues on our l10n-spanish list. I wanted to people directly learn and use the tools we do and also the infrastructure, so at some point I made the decision of make them directly work on our team's infra, this of course got some people inquiring if I was running a parallel project and complain about the quality, later I had to explain what was this rush of translations and that we were running this workshop and, yes, I knew the work was not going to be as we usually expect but they were quite new and we didn't have much time for have them learn all our policy, terms and all things one knows after being on the team for a while.
My girlfriend also involved herself and did some translation work and took pictures for me, cute. However, since I had an horrible issue with my main domain for email, and also because my time constraints, I didn't have much time to follow them up but I'm glad about what they did and that some of them are very encouraged and still working on the list. One of our goals was to get at least one people "hooked" on the direct contribution I'm glad to see we over-passed this goal. The report has the in-depth details, like number of participants, talks, also the problems we faced, as Diego describes (no, don't believe him, I wasn't acting as the contact/manager guy all the time, word). We had a dinner later with some of the people who still had energies or were hungry ;).
There is a pictures gallery at our website, also the talks slides, the videos are still on processing. We'll publish them once they are ready, like everything in Debian of course ;). Finally, I want to say that this was a great personal experience to transmit the things I've learnt and having people to feel that they also can do it, but even more: doing it!. Sometimes online chat or mailing lists are quite cold for transmitting this knowledge to newcomers. Usually they got scared after one tells them they have to read a lot, or they just got frustrated by their inability to understand the team work-flow at the very beginning. I'd like to thank our wonderful band, or squad (you named it, we can get it for you), who is responsible for making this happen. We are already working on the long-term (call strategic if you like it) plan for the up-comings invol crate+. We've already made a shot-term plan for this year and started to work on the next invol crate+, be ready.
Of course thanks to Peru Pacifico Institute where we hosted the event. We had a post-event meeting with them and directors are very happy with us and with the success, they are willing to work together again on the future. Also to our helpful sponsors, speakers, friends and relatives who support us.

There aren’t many features I grew fond of with CVS that I haven’t yet found a satisfactory solution for in Darcs. One of them was, until today, sending commit mails to mailing lists. There are a number of more or less ugly hacks floating around the net, but I never found them very satisfactory, so I wrote my own. Darcs-monitor will send email to a specified recipient about new changes added to a specific darcs repository. It can be run as an apply posthook (resulting in near-instantaneous “push” emails), or periodically from Cron, or occasionally by hand, whatever seems most convenient. You can get it by darcs at http://antti-juhani.kaijanaho.fi/darcs/darcs-monitor/ (darcsweb available), or as a tarball at http://antti-juhani.kaijanaho.fi/software/dist/. It depends on mtl and HaXml and is written in Haskell, naturally. It has been tested only with GHC 6.6 so far. For installation and usage instructions, refer to the README. Please note that this software is very new and thus can be buggy. As with any email-sending software, bugs can result in major annoyance; user caution is warranted.

Have you ever wondered how you can easily capture a console session on Linux (and probably most other Unix-like OSes) without the need for special video recording software? Here's how: you can use the script(1) command (part of the bsdutils package).

• Start the script command. Without parameters the sessions will be logged to a file called typescript. You can also supply a filename on the command line: script mysession.log.
• Type whatever you want to record...
• End the recording session with exit or CTRL-d.
• You can now view the file with less -r mysession.log (it contains line feeds, escape sequences etc., so a simple text viewer will probably display garbage). You can even print it with lpr(1). Or simply display it on the terminal with cat mysession.log.

An even cooler feature is that you can replay a script with the correct timing information:

• First you have to also record the timing information using the -t switch: script -t 2>mysession.timing mysession.log.
• You can then replay the script: scriptreplay mysession.timing mysession.log.

See the script(1) and scriptreplay(1) manpages for more details.

Ulrich Drepper has written a good web page about text relocation which is most often noticed as execmod failures reported when running SE Linux. When an AVC message reports a failure of execmod against a shared object it means that the object has text relocations (the shared object code writes to code that it executes to fix up addresses). This is due to being compiled without -fPIC or -fpic.

The command eu-findtextrel (from the elfutils package) when run with a parameter of the shared object in question will tell you which functions were compiled without -fpic or -fPIC.

The module in question must be recompiled with -fpic or -fPIC to generate the correct code.

Without SE Linux it's still a bug to compile a shared object without position independent code, so any shared object which can't run under SE Linux because of execmod will probably have problems in other situations anyway (maybe only on certain architectures).

One of the benefits of Xen is that it allows a machine to be easily rebooted. Remote console and remote power management technologies are either expensive or implemented on the motherboards of expensive machines. With Xen the virtual machines can be managed without such expense and also with less effort.

This raises immediate possibilities for training sys-admins. One problem with training system administrators is that they need to have servers to administer and the mentor needs to be able to easily access them and fix them when they become unbootable. Xen makes these problems easy to solve on cheap hardware.

Monash University has for many years had a machine named Yoyo which is used for training sys-admins. It's not expected to be up as much as machines that are run by experienced sys-admins (hence the name) but I expect it would still have better uptime than a lot of corporate servers.

Unfortunately until recently there were no options available to people who weren't Monash students for learning about system administration. To solve this I want to run some sys-admin training with a server.

My plan is to set up a machine running Xen at a server room and provide a basic Debian install in a domU. I will then give the root password to a small group of trustworthy people (who I have known for some time or who are local and can be verified) and start the training. My plan is to set up a mailing list for emergency communication on one of my servers and have the trainees set one up for regular use on the domU. I will provide DNS secondary service and have an NS record for yoyo.coker.com.au point at the machine in question so the first task for them would be to run a DNS server.

I won't have unlimited bandwidth so I will track the bandwidth use from the dom0, and I will use the dom0 to make backups of the system via LVM snapshots.

The terms of service etc will be mostly copied from the Monash machine.

Please let me know if you have any suggestions for how to run this.

Hopefully this will work out and other people will want to do the same. If you run such a machine then please add a comment to this post with the URL for information on it.

I've been taking the train frequently lately so decided to sign up for the Amtrak Guest Rewards Program -- kind of like a frequently flyer program but without the flying. When filling out their online join form, I was asked to select a title from what turned out to be a rather extensive list:

Please Select Admiral Airman Brother Captain, USA Captain, USN Chief Master Sgt. Chief Petty Officer Chief Warrant Officer 4 Col. Commander Dr. Father Gen. Hon. Jurist Doctorate Lady Lieutenant Commander Lt. Lt. Col. Lt. Gen. Maj. Gen. Major Master Chief Master Sgt. Miss Mother Mr. Mrs. Ms. Pastor Prince Princess Prof. Rabbi Rev. Father Reverend SIR Senator Senior Chief Sgt. Sgt. Maj. Sister

That will be Princess Benjamin Mako Hill to you, Mr. Conductor.

... suck. The XML Schema "datetime" format can't be handled by Java's SimpleFormat (ok, that probably is more Javas fault, of not being able to handle hours:minutes in the timezone specification, anyway, this is mildy annoying). The XSLT2 parsers are very intolerant about the format of the time specification, too. They could have made more stuff optional such as the specification of seconds; an error here will make the whole XSLT fail with an exception. Some compact form of error handling would be nice... as would be a smart parser which can handle various formats. The XML Schema "duration" is even worse. First of all, it was completely forgotton when doing XSLT 2; there are no functions to format or disassemble it (except by regular expressions, which could also use a zero-width lookahead). Secondly, it's lacking common specifications such as "next week". While "next week" is computationally equivalent to "in 7 days", it can have different semantics in some contexts (especially when not being aligned): If I'm looking for the week February 9th 2007 is in, the result is February 5th two February 11th. If I'm looking at a 7 day interval containing this day, there are infinite possibilities (aligned on milliseconds and below...). So it does make sense to make a difference between 7 days and a week.