Since some ages we are deploying IPv6 in our network and also for some selected services. Some days ago we discovered, that anybody has enabled accidentally a router advertisement daemon in a network where this shouldn t happen. As result of this, IPv6 enabled systems got (additional) IPv6 adresses and some services where using this new learned addresses as source address when sending out replies. Maybe this doesn t sound so harmful on the first view. But some services rely on a correct source address when getting a reply for a request (for example dns resolver library). To be aware of the issue, that a service maybe available via IPv4 but not via IPv6 or vica versa, a dualstacked monitoring needs to be in place for dualstacked services. Looking into the issue, I stumbled upon Michael Friedrichs HowTo about Dualstacked monitoring with Icinga. Luckily we have the full software stack in place on Squeeze (with squeeze-backport). As icinga and nagios-plugins was in installed already, I just needed to fetch check_multi.

aptitude install -t squeeze-backports nagios-plugin-check-multi Now a new check command is needed, I created the following:

define command
command_name check_multi_icinga
command_line /usr/lib/nagios/plugins/check_multi \
-f /etc/check_multi/$ARG1$ $ARG2$ $ARG3$ $ARG4$ \
-s objects_cache=/var/cache/icinga/objects.cache \
-s status_dat=/var/cache/icinga/status.dat \
-s HOSTADDRESS=$HOSTADDRESS$ -s HOSTADDRESS6=$HOSTADDRESS6$
For monitoring connectivity I created check_host_alive_dualstack.cmd

command[ IPv4 ] = check_ping -4 -H $HOSTADDRESS$ -w 5000.0,100% -c 5000.0,100% -p 5
command[ IPv6 ] = check_ping -6 -H $HOSTADDRESS6$ -w 5000.0,100% -c 5000.0,100% -p 5
state [ CRITICAL ] = COUNT(CRITICAL) > 1
state [ WARNING ] = COUNT(WARNING) > 0 COUNT(CRITICAL) > 0
state [ UNKNOWN ] = COUNT(UNKNOWN) > 1 Now we just need to replace the check-host-alive command and add a value for address6 on the host as the following

define host
use generic-host ; Name of host template to use
host_name localhost
alias localhost
address 127.0.0.1
address6 ::1
check_command check_multi_icinga! check_host_alive_dualstack.cmd !'-r 1+2+4+8
Reloading icinga should you show something like this:
Now we have general connectivity of our dualstacked systems monitored.

Inspired by the interview series conducted by Raphael, I started a Norwegian interview series with people involved in the Debian Edu / Skolelinux community. This was so popular that I believe it is time to move to a more international audience. While Debian Edu and Skolelinux originated in France and Norway, and have most users in Europe, there are users all around the globe. One of those far away from me is Nigel Barker, a long time Debian Edu system administrator and contributor. It is thanks to him that Debian Edu is adjusted to work out of the box in Japan. I got him to answer a few questions, and am happy to share the response with you. :) Who are you, and how do you spend your days? My name is Nigel Barker, and I am British. I am married to Yumiko, and we have three lovely children, aged 15, 14 and 4(!) I am the IT Coordinator at Hiroshima International School, Japan. I am also a teacher, and in fact I spend most of my day teaching Mathematics, Science, IT, and Chemistry. I was originally a Chemistry teacher, but I have always had an interest in computers. Another teacher teaches primary school IT, but apart from that I am the only computer person, so that means I am the network manager, technician and webmaster, also, and I help people with their computer problems. I teach python to beginners in an after-school club. I am way too busy, so I really appreciate the simplicity of Skolelinux. How did you get in contact with the Skolelinux/Debian Edu project? In around 2004 or 5 I discovered the ltsp project, and set up a server in the IT lab. I wanted some way to connect it to our central samba server, which I was also quite poor at configuring. I discovered Edubuntu when it came out, but it didn't really improve my setup. I did various desperate searches for things like "school Linux server" and ended up in a document called "Drift" something or other. Reading there it became clear that Skolelinux was going to solve all my problems in one go. I was very excited, but apprehensive, because my previous attempts to install Debian had ended in failure (I used Mandrake for everything - ltsp, samba, apache, mail, ns...). I downloaded a beta version, had some problems, so subscribed to the Debian Edu list for help. I have remained subscribed ever since, and my school has run a Skolelinux network since Sarge. What do you see as the advantages of Skolelinux/Debian Edu? For me the integrated setup. This is not just the server, or the workstation, or the ltsp. Its all of them, and its all configured ready to go. I read somewhere in the early documentation that it is designed to be setup and managed by the Maths or Science teacher, who doesn't necessarily know much about computers, in a small Norwegian school. That describes me perfectly if you replace Norway with Japan. What do you see as the disadvantages of Skolelinux/Debian Edu? The desktop is fairly plain. If you compare it with Edubuntu, who have fun themes for children, or with distributions such as Mint, who make the desktop beautiful. They create a good impression on people who don't need to understand how to use any of it, but who might be important to the school. School administrators or directors, for instance, or parents. Even kids. Debian itself usually has ugly default theme settings. It was my dream a few years back that some kind of integration would allow Edubuntu to do the desktop stuff and Debian Edu the servers, but now I realise how impossible that is. A second disadvantage is that if something goes wrong, or you need to customise something, then suddenly the level of expertise required multiplies. For example, backup wasn't working properly in Lenny. It took me ages to learn how to set up my own server to do rsync backups. I am afraid of anything to do with ldap, but perhaps Gosa will help. Which free software do you use daily? Nowadays I only use Debian on my personal computers. I have one for studio work (I play guitar and write songs), running AV Linux (customised Debian) a netbook running Squeeze, and a bigger laptop still running Skolelinux Lenny workstation. I have a Tjener in my house, that's very useful for the family photos and music. At school the students only use Skolelinux. (Some teachers and the office still have windows). So that means we only use free software all day every day. Open office, The GIMP, Firefox/Iceweasel, VLC and Audacity are installed on every computer in school, irrespective of OS. We also have Koha on Debian for the library, and Apache, Moodle, b2evolution and Etomite on Debian for the www. The firewall is Untangle. Which strategy do you believe is the right one to use to get schools to use free software? Current trends are in our favour. Open source is big in industry, and ordinary people have heard of it. The spread of Android and the popularity of Apple have helped to weaken the impression that you have to have Microsoft on everything. People complain to me much less about file formats and Word than they did 5 years ago. The Edu aspect is also a selling point. This is all customised for schools. Where is the Windows-edu, or the Mac-edu? But of course the main attraction is budget.The trick is to convince people that the quality is not compromised when you stop paying and use free software instead. That is one reason why I say the desktop experience is a weakness. People are not impressed when their USB drive doesn't work, or their browser doesn't play flash, for example.

Yesterday, I've finalized Weblate 0.4. It improved handling of plural strings, added documentation and implemented support for secondary languages. If you ask for 0.3, well I managed to release two versions in the day, but I expect to slow down this pace. Full list of changes for 0.3:

• Better display of source for plural translations.
• New documentation in Sphinx format.
• Displays secondary languages while translating.
• Improved error page to give list of existing projects.
• New per language stats.

Full list of changes for 0.4:

• Added usage guide to documentation.
• Fixed API hooks not to require CSRF protection.

You can find more information about Weblate on it's website, the code is hosted on Github. If you are curious how it looks, you can try it out on demo server. You can login there with demo account using demo password or register your own user. Weblate is now also being used https://l10n.cihar.com/ as official translating service for phpMyAdmin, Gammu and others.

Filed under: English Weblate 0 comments Flattr this!

FedEx has had some issues in my area. There for awhile, they kept returning packages as undeliverable, saying the address was bad. It wasn t. Just because their GPS can t find it or their driver doesn t know it doesn t mean it doesn t exist. A map, a real map, would have done it. I had this kind of conversation with them once. Me: So you re returning my package to the sender. FedEx: Yes sir, the address was wrong. Me: No, it wasn t. [ we confirm we're talking about the same address. ] FedEx: Our driver says it doesn t exist. Me: I ve lived here for [x] years. I do know my address. FedEx: Maybe it s just new. Me: Someone has lived at this spot since the 1890s. FedEx: Hmm. I don t know then. Me: How about I give you directions? FedEx: Well, OK, I guess I could send directions. But they probably won t see them for three days. Me: Where will my package be by then? FedEx: Hard to say. Me: Do you people have cell phones maybe? FedEx: No. Me: Well, OK then Lately we have been noticing that the driver throws the package from the truck without stopping. We find them all over the yard. Unless it s big, in which case he puts it on the porch, puts a completely useless sticker on the door saying We delivered your package to right here! , and doesn t ring the doorbell. And panics due to the muddy road. The FedEx driver once told Terah, I almost got stuck on that muddy road! What s the good one? Terah: Well, which one did you take? [ he describes ] Terah: [ serious ] That IS the good road. FedEx: [ turning pale ] Oh my. I don t know if I can make it back to the highway. Terah: Well, I suppose I could get the boys in the car and drive you into town FedEx: [ embarrassed and more pale ] Uhm that s OK uhm I think A *month* later, we got another FedEx package. The driver called me at work, saying, I just don t know how I m going to get down those roads. Can I deliver the package to you here? (10 miles away). I said yes, and managed to suppress a significant snicker at his obvious relief. FedEx drivers seem to come and go. The UPS driver, on the other hand, knows the names of our boys, our cat, and puts anything small behind the storm door so it won t blow away if we re not home. And he gets a great big laugh when I tell him that the FedEx drivers are afraid of our roads. Haha, he thinks THAT S mud? Obviously he never drives in Butler County.

Hey I'm alive. I was jet lagged, and then caught a flu and wasn't active for the few weeks, but now back.

I followed Gregor s evolution within Debian because I used to be somewhat active in the Perl team. His case is exemplar because it shows that you don t need to be an IT professional to join Debian and to make a difference. His QA page is impressive with hundreds of packages maintained and hundreds of non-maintainer uploads too. While he started out slowly, I remember meeting him at Debconf 7 in Edinburgh and after that he really got more implicated. Again a case of someone joining for technical reasons but getting more involved and staying there for social reasons! Let s jump into the interview and learn more about him. Raphael: Who are you? Gregor: I m 41 years old, and I live in Innsbruck, Austria, in a shared apartment with a friend of mine. In my day job, I m working at the regional addiction prevention agency, so I m one of the few Debian guys who s not an IT student or professional. I started maintaining packages in 2006, and I am a DD since April 2008. Raphael: How did you start contributing to Debian? Gregor: After having used Debian on servers for some years, I finally switched to it on the desktop after some procrastinating. Soon afterwards I wanted to know more about the making-of , started to join mailing lists, filed bugs, and tried to learn packaging. Luckily I quickly found a permanent sponsor Tony Mancill, and we re still co-maintaining each others packages. And when I packaged my first Perl modules, Gunnar Wolf invited me to join the Debian Perl Group, an offer I accepted a few days later. And I m still there Later, the NM process, although it involved some waiting times, was also a good learning experience due to my AM Wouter Verhelst. (And in the meantime the organization of the NM process has vastly improved, from what I hear.) So my starting point for joining Debian was my curiosity but what really helped me to find my way into the project was the support of the people who invited and helped me. Raphael: What s your biggest achievement within Debian or Ubuntu? Gregor: I m not sure I can name a single big achievement but I guess I can say that my contributions to the Debian Perl Group have helped to make and keep the team a success story. Raphael: The pkg-perl team seems to work very well. As an active member, can you explain us how it is organized? How do you explain its success? In particular it seems to be a great entry point for new contributors. Gregor: The team is huge, both in numbers of members and packages (over 2200). Since last DebConf we manage our source packages in git, we have 2 mailing lists and an IRC channel, and we manage to keep an overview by using PET, the Package Entropy Tracker. It s true that we get new members on a regular basis; we try to invite people (like it happened to me 6 years ago ) but there are also quite a few new contributors who find our docs and introduce themselves on the mailing list. Maybe someone should conduct a study and ask them what motivated them to join. We hand out group membership/commit access quickly, and we try to mentor new contributors actively during their early times in the group. Some of them leave for other projects after some time, but many also stay and become DDs later. I m not sure what the reasons for the group s success are, maybe a combination of:

• a culture in the group (and also in the upstream Perl community) that s based on fun, cooperation, and respect;
• the fact that packaging Perl modules is most of the time quite easy;
• a great set of tools, and also (hopefully) useful documentation;
• a bunch of relaxed people who are open to newcomers and try to help each other.

For everyone interested in joining the Debian Perl Group, our Welcome page on the wiki is a good starting point. Raphael: What are your plans for Debian Wheezy? Gregor: Nothing overly exciting. What I should do is getting a newer JabRef into Debian (which involves packaging some new Java libraries any takers?). A solution for libdatetime-timezone-perl (which ships timezone data converted to Perl modules and tends to get outdated when the timezone data change) would be nice; let s see if #660404 leads to some results And some Perl packages will also need a bit of work for the hardening build flags release goal (cf. #657853). Raphael: What s the biggest problem of Debian? Gregor: Inertia. While I really like the fact that Debian is a volunteer project, and that every contributor works when and on what they decide to work on, I get the feeling that Debian could do better in moving forward, innovating, taking decisions. I also think that more uniformity in managing source packages would make things easier; it s quite amazing to see how many source formats, packaging helpers, patch systems, RCSs etc. are used all over the archive. I m not advocating for mono-cultures, and I consider this diversity a strength in general, but having to find out first how this particular package works when preparing a bug fix can be annoying. On the bright side, I think that the myth Debian and its mailing lists are mostly about flames can be seen as dispelled in the meantime. Sure, sometimes the tone could be a bit more civil, but in general most of the interactions I ve seen in the last years were friendly and helpful. IMO, the Debian Project consists of mostly nice and cooperative people, and that s what makes it fun for me. Raphael: You re one of the most dedicated participants to RCBW (Release Critical Bugs of the Week), an initiative to fix RC bugs every week. How much time do you spend on it? What would you advise to people who are considering to join the movement? Gregor: I got into the habit of fixing RC bugs after having been invited to my first Bug Squashing Party in Munich some years ago. During this weekend I saw that fixing RC bugs can be fun, is often not that difficult, and gives a warm fuzzy feeling I can definitely recommend attending a BSP if one happens to be organized near you. After tasting blood at this first BSP I tried to continue looking at RC bugs, and I guess I spend something around half an hour per day on it. I usually blog about it once a week, in order to motivate others to join in. And joining is easy: just take a look at the tips people like Zack, Vorlon, or me have written. You don t have to be a DD to help, many of my NMUs are based on patches that others kindly prepare and send to the BTS kudos! Another nice aspect is that the RC bug list contains problems from different fields: general packaging problems, language-specific issues, policy violations, etc. So there s something for everybody, and you don t have to be an expert in all fields to fix a specific bug. What s rewarding about fixing RC bugs is not only the feeling of accomplishment and the knowledge about having helped the next release I also received quite a few Thank you mails from maintainers who were busy at that time and appreciated the help. Raphael: Do you have wishes for Debian Wheezy? Gregor: Well, there s not so much left of the Wheezy release cycle if we manage to freeze in June Some quick thoughts for Wheezy and Wheezy+1:

• Obviously, getting multi-arch and the hardening build flags as far as possible would be good.
• What I like is the idea of the time-based freeze, and I hope it will work out in June. And then I hope that the freeze will be shorter this time than during the last 2 releases.
• Unless I m missing something, the CUT discussions have more or less died down; IMO that s a pity because there are users who run testing and would like to avoid the several month long freeze. Maybe someone can come up with new ideas for Wheezy+1
• Too late for broad adoption in Wheezy but still: What constantly annoys me is the handling of conffiles during upgrades (when I want to keep changed values but at the same time want to add new variables). Config::Model seems to be the best idea so far for configuration upgrades but it s not yet widely adopted.

Raphael: Is there someone in Debian that you admire for their contributions? Gregor: There are many people in Debian I admire, too many to name them all. The first one that comes to my mind is Russ Allbery who not only does great work from lintian to Debian policy but who also sets a great example of communicating in a perfectly polite and respectful way even in heated discussions.

---

Thank you to Gregor for the time spent answering my questions. I hope you enjoyed reading his answers as I did. Note that older interviews are indexed on wiki.debian.org/PeopleBehindDebian.

Subscribe to my newsletter to get my monthly summary of the Debian/Ubuntu news and to not miss further interviews. You can also follow along on Identi.ca, Google+, Twitter and Facebook.

2 comments Liked this article? Click here. My blog is Flattr-enabled.

I ll be at PyCon with some of the Sunlight crew, and we ll be hosting a Hackathon after the talks. I hope anyone in the area will consider hacking with us this year. If you re going to PyCon, send me an email, tweet or other digital message - the talks this year look awesome!

Thanks to klibc being fixed, rootskel finally built in the archive and hence we've now finally enabled the daily builds of debian-installer for s390x. They're still untested, though, and I hope to come around to that in the near future.

In other news I've spent some more time chasing weird 64bit big endian issues in glib. Newer versions have regressed in their support and again assume that certain fields are either 32bit/64bit little endian or 32bit big endian, which is unhelpful. Sadly the testsuite is guilty as well, which doesn't make debugging any easier. I still suspect a bug in either gio or GClosure's interface to libffi, let's hope that when that one's found that the remainder of the archive is building just fine (Currently a lot is blocking on glib-networking which fails in its testsuite. And of course there are still usertagged bugs that need to be fixed.)

It would be cool if we could run more testsuites during package building and find the bugs in them. glib does have one, but its failures are non-fatal for the build (also because there are so many failures). That would make porting to future architectures a tad easier.

Packt Publishing has offered phpMyAdmin to sponsor some coding contest with books. The contest is already running since Monday and we have first winner, but still, your chances are pretty high. So in case you are interested in having Mastering phpMyAdmin 3.4 for Effective MySQL Management, go ahead and check out the contest rules.

Filed under: English Phpmyadmin 0 comments Flattr this!

RcppArmadillo release 0..2.36 is now on CRAN. It contains just the changes from the new Armadillo release 2.4.4. The NEWS entry below summarises the changes.

0.2.36  2012-03-05
    o   Upgraded to Armadillo release 2.4.4
          * fixes for qr() and syl()
          * more portable wall_clock class
          * faster relational operators on submatrices

Courtesy of CRANberries, there is also a diffstat reports for 0.2.36 relative to 0.2.35 As always, more detailed information is on the RcppArmadillo page. Questions, comments etc should go to the rcpp-devel mailing list off the R-Forge page.

I'd like to setup some auto-builders for some projects - and theese projects must be built upon Lenny, Squeeze, Lucid, and multiple other distros. (i386 and amd64 obviously.) Looking around I figure it should be simple. There are a lot of continuous integration tools out there - but when looking at them in depth it seems like they all work in temporary directories and are a little different to how I'd expect them to be. Ultimately I want to point a tool at a repository (mercurial), and receive a status report and a bunch of .deb packages for a number of distributions. The alternative seems to be to write a simple queue submission system, then for each job popped from the queue run:

• Creates a new debootstrap-based chroot.
• Installs build-essential, mercurial, etc.
• Fetches the shource.
• Runs make.
• Copies the files produced in ./binary-out/ to a safe location.
• Cleans up.

Surely this wheel must already exist? I guess its a given that we have to find build-dependencies, and that we cannot just run "pbuilder *.dsc" - as the dsc doesn't exist in advance. We really need to run "make dependencies test build", or similar. Hudson looked promising, but it builds things into /var/lib/hudson, and doesn't seem to support the use of either chroots or schroots. ObQuote: "I feel like I should get you another sweater." - "Friends"

As the free software and free culture movements have sat quietly by, DRM is now well on its way to becoming the norm in the electronic book publishing industry. The free culture movement has failed to communicate the reality of DRM and, as a result, millions of people are buying books that they won't be able to read when they switch to a different model of ebook reader in the future. They are buying books that will become inaccessible when the DRM system that supports them is shut down -- as we've already seen with music from companies including Wal*Mart, Yahoo, and Microsoft. They are buying books that require that readers use proprietary tools that lock them out from doing basic things that have always been the right of a book owner. Some anti-DRM advocates are, indirectly, part of this problem as they buy these books and turn to shady methods of stripping the DRM. Buying DRMed books is voting with your wallet for a system that criminalizes those that insist on living in freedom and will screw us all in the long run when DRM is the only choice we are offered and removing the DRM is difficult, unsafe, and illegal. Buying non-DRMed e-books is a more freedom-friendly alternative for those that, like me, are excited about not lugging kilograms of paper around our cities and the world. We can do this at "non-mainstream" publishers like Smashwords who explicitly reject DRM. Of course, the big ebook sellers like Amazon, and Barnes and Nobel, and Google all offer non-DRMed books. But none of the major ebook retailers explicitly reveal the DRM status of locked down books before purchase. On Amazon, there are some cryptic signs and signals that, if you understand them, suggest the absence of DRM. Google and Barnes and Nobel currently offer no way to know if a book is DRMed without buying it first and questions in their support forums go unanswered. It's hard to support non-DRM alternatives when we can't recognize them. It's hard to tell people to not buy DRM ebooks if we can't even tell them apart. Getting this message through to book buyers -- and perhaps even to ebook retailers -- seems like a critical first step.

After about week of (not so heavy) development, I can announce availability of Weblate - web based translation tool with tight Git integration. The project was started because I was not happy with existing solutions (see my blog post) as all of them was missing some features I find useful. I started with quite simple definition:

Minimalistic web based translation with direct commit to git on each translation made. There is no plan in heavy conflict resolution as these should be primarily handled on git side.

Also the project started with defined list of features:

• Easy web based translation
• Propagation of translations across sub-projects (for different branches)
• Tight git integration
• Usage of Django's admin interface
• Upload and automatic merging of po files
• Links to source files for context

Pretty much all of that is completed right now, however there will be definitely some issues which will need fixing. All important information about the project are available on it's website, the code is hosted on Github. If you are curious how it looks, you can try it out on demo server. You can login there with demo account using demo password or register your own user. Please give it some testing as I plan to deploy it on http://l10n.cihar.com/ pretty soon.

Filed under: Django English Weblate 5 comments Flattr this!

I recently got to rebuild an Android NDK with a fresh toolchain again, and hit an interesting problem. I actually had hit it before, but only this time I fully analyzed what s going on. [As a side note, if you build such a NDK, don't use mpfr 3.1.0, as there is a bug in the libtool it ships] Linking an application or a library pulls many things, that aren t part of the code being built. One of these many things is the libgcc static library. Part of libgcc consists in an implementation of the platform ABI. On Android systems, this means the ARM EABI. GCC, when compiling some instructions, will generate ABI calls. For example, integer divisions may call __aeabi_idiv. Consider the following minimized real world scenario:

$ echo "int foo(int a) return 42 % a; " > foo.c
$ arm-linux-androideabi-gcc -o libfoo.so -shared foo.c -mandroid

GCC will emit a call to __aeabi_idivmod for the % operation. With GCC 4.6.3, this function is in _divsi3.o under libgcc.a. That function itself calls __aeabi_idiv0, which lives in _dvmd_lnx.o under libgcc.a. When statically linking, ld will thus include foo.o, _divsi3.o and _dvmd_lnx.o, meaning it will include all functions from these object files. That is, foo, __divsi3, __aeabi_idiv, __aeabi_idivmod, __aeabi_idiv0 and __aeabi_ldiv0. And more than being included, these functions are exported, because symbol visibility in libgcc.a is default. So while we expect exporting foo from our library, we re actually exporting much more, including functions that just happened to be near the ones that our code (indirectly) uses. Now, let s say we want to build another library, using that foo function from libfoo:

$ cat > bar.c <<EOF
extern int foo(int a);
long long bar(long long a) return foo(a) % a;
EOF
$ arm-linux-androideabi-gcc -o libbar.so -shared bar.c -mandroid

(The code above has absolutely no meaning, it just triggers the same function calls as what I was getting in the actual real world case) When statically linking the above code, GCC will generate a call to __aeabi_ldivmod, which calls __aeabi_ldiv0, and many other things, directly or indirectly. When linking as above, nothing particularly nasty is going to happen. However, linking as above is actually wrong: the resulting library has an undefined reference to the foo symbol, and doesn t depend on libfoo. At runtime, if libfoo wasn t already loaded somehow, loading libbar would fail. The proper way to link is the following:

$ arm-linux-androideabi-gcc -o libbar.so -shared bar.c -mandroid -L. -lfoo

A feature of ELF static linking is that when it resolves undefined symbols, the linker will choose to use the first occurrence of a symbol it finds in the various objects and libraries given on its command line. So with the command line above, for each __aeabi_* symbol, it will first look in libfoo if there isn t one. And while __aeabi_ldivmod is not in libfoo, __aeabi_ldiv0 is (see above). So instead of including the code for __aeabi_ldiv0 from libgcc.a, it will call the copy from libfoo. This wouldn t be so much of a problem if __aeabi_ldiv0 wasn t a weak symbol. Enters faulty.lib. In the real world case, libfoo is loaded by the system dynamic linker, and libbar by faulty.lib. When resolving symbols for libbar, faulty.lib has to resolve libfoo symbols with the system linker, using dlsym(). On Android, dlsym() returns NULL for weak (defined) symbols, so faulty.lib can t resolve __aeabi_ldiv0. The real world case wasn t a problem with GCC 4.4.3 from the vanilla Android NDK because in that GCC version, __aeabi_ldivmod doesn t call __aeabi_ldiv0. This wouldn t happen if shared libraries wouldn t expose random platform ABI specific bits depending on what they use and depending on other symbols that happen to be in the same object files. A similar issue happened a little while ago on Debian powerpc because a shared library was exporting ABI specific bits. Even worse, the toolchain was assuming the symbols would come from libgcc.a and generated wrong relocations for these symbols. Update: Interestingly, the __aeabi_* symbols are hidden, in libgcc.a as provided on the Debian armel port.

* We ve now done with big transition of Indic fonts from ttf-* naming to fonts-*. Debian-IN-Workers team now maintains around 61 packages. We ve also finished process of moving to Git from SVN along with this. Kudos to Vasudev, Muneeb and AbdulKarim for taking lead, help and pushing me hard for upload and testing. (31 packages upload on Sunday night = 62 times my GPG passphrase, not counting couple of wrong entries )

You work with Python for a while and you'll become familiar with printing a method and getting I think there is room for one more explanation on the internet, since I've never seen it diagrammed out (maybe for good reason!). In the above diagram on the left, we have the fairly simple conceptual model of a class with a function. One naturally tends to think of the function as a part of the class and your instance calls into that function. This is conceptually correct, but a little abstracted from what's actually happening. The right attempts to illustrate the underlying process in some more depth. The first step, on the top right, is building something like the following: As this illustrates, the above code results in two things happening; firstly a function object for function is created and secondly the __dict__ attribute of the class is given a key function that points to this function object. Now the thing about this function object is that it implements the descriptor protocol. In short, if an object implements a __get__ function; then when that object is accessed as an attribute of an object the __get__ function is called. You can read up on the descriptor protocol, but the important part to remember is that it passes in the context from which it is called; that is the object that is calling the function. So, for example, when we then do the following: what happens is that we get the attribute function of f and then call it. f above doesn't actually know anything about function as such what it does know is its class inheritance and so Python goes searching the parent's class __dict__ to try and find the function attribute. It finds this, and as per the descriptor protocol when the attribute is accessed it calls the __get__ function of the underlying function object. What happens now is that the function's __get__ method returns essentially a wrapper object that stores the information to bind the function to the object. This wrapper object is of type types.MethodType and you can see it stores some important attributes in the object im_func which is the function to call, and im_self which is the object who called it. Passing the object through to im_self is how function gets it's first self argument (the calling object). So when you print the value of f.function() you see it report itself as a bound method. So hopefully this illustrates that a bound method is a just a special object that knows how to call an underlying function with context about the object that's calling it. To try and make this a little more concrete, consider the following program: Running this gives output something like To pull it apart; we can see that Foo.__dict__['function'] is a function object, but then f.function is a bound method. The bound method's im_func is the underlying function object, and the im_self is the object f: thus im_func(im_self) is calling function with the correct object as the first argument self. So the main point is to kind of shift thinking about a function as some particular intrinsic part of a class, but rather as a separate object abstracted from the class that gets bound into an instance as required. The class is in some ways a template and namespacing tool to allow you to find the right function objects; it doesn't actually implement the functions as such. There is plenty more information if you search for "descriptor protocol" and Python binding rules and lots of advanced tricks you can play. But hopefully this is a useful introduction to get an initial handle on what's going on!

One of the best way to add SNMP support to an application is the use of AgentX protocol: an application can act as a subagent and handle requests from a master agent using AgentX protocol. The most widespread implementation for such a task is the one from Net-SNMP. TL;DR: An application asynchronously leveraging Net-SNMP AgentX implementation has many opportunities to get stuck when the master agent becomes temporarily unavailable. A mostly efficient workaround is to dedicate a master agent for AgentX and delegate MIB handling to a subagent.

Pinging the master agent Net-SNMP uses an event-based model. It also provides some synchronous functions but they are built on top of the asynchronous ones. It is shipped with its own event loop. Therefore, Net-SNMP AgentX protocol implementation seems a fine choice to add SNMP support to an event-based program, for example in Keepalived, a VRRP daemon and a monitoring daemon for LVS clusters. Unfortunately, an AgentX subagent is expected to check, at a regular interval, if the master agent is alive . This is handled automatically with Net-SNMP implementation as long as you call run_alarms() function which will call agentx_check_session():

/*
 * check a session validity for connectivity to the master agent.  If
 * not functioning, close and start attempts to reopen the session 
 */
void
agentx_check_session(unsigned int clientreg, void *clientarg)
 
    /* [...] */
    DEBUGMSGTL(("agentx/subagent", "checking status of session %p\n", ss));
    if (!agentx_send_ping(ss))  
        snmp_log(LOG_WARNING,
                 "AgentX master agent failed to respond to ping.  Attempting to re-register.\n");
        /*
         * master agent disappeared?  Try and re-register.
         * close first, just to be sure .
         */
        agentx_unregister_callbacks(ss);
        agentx_close_session(ss, AGENTX_CLOSE_TIMEOUT);
        /* [...] */
        if (main_session != NULL)  
            /* [...] */
            main_session = NULL;
            agentx_reopen_session(0, NULL);
         
        else  
            snmp_close(main_session);
            main_session = NULL;
         
      else  
        DEBUGMSGTL(("agentx/subagent", "session %p responded to ping\n",
                    ss));
     
 

This piece of code will end up calling (directly or indirectly) the following functions¹ which are using agentx_synch_response():

• agentx_send_ping()
• agentx_open_session()
• agentx_reopen_session()
• agentx_close_session()
• agentx_register()
• agentx_unregister()
• agentx_register_index()
• agentx_unregister_index()
• agentx_add_agentcaps()
• agentx_remove_agentcaps()

agentx_synch_response() will synchronously wait for an answer from the master agent. Your event-based program will just sit here doing nothing while this happens. For example, keepalived may be unable to send VRRP probes resulting in some serious havoc in your cluster. Functions prefixed by agentx_ are not part of NetSNMP API and could therefore be rewritten to do their work asynchronously. Unfortunately, talk is cheap. Moreover, there exists other deep paths that will block a subagent. For example, the low-level connection to the master agent (using TCP or a Unix socket, but not with UDP) will use a blocking connect() call. It seems quite difficult to fix this. Here are some (partial) workarounds:

• Call agentx_check_session() less often. The default interval is 15 seconds. Disabling it completely or using a very high value is not advised because this function is the only way to reestablish a connection to the master agent when it is, for example, restarted. Here is how to change the delay to 120 seconds:

  netsnmp_ds_set_int(NETSNMP_DS_APPLICATION_ID,
                     NETSNMP_DS_AGENT_AGENTX_PING_INTERVAL, 120);
  

• Don t call agentx_check_session() if we know the master agent is alive. For example, if it just sent us a new request, no need to ping it. I have written a patch implementing such a workaround. The ABI is left untouched, therefore, you only need to recompile Net-SNMP libraries.
• Use a lower timeout and no retry. By default, the timeout for synchronous requests is 1 second and a failed request will be retried 5 times. Because several synchronous functions are used in a row, you may be trapped for as long as 30 seconds with those settings. Here is how² to disable the retry mechanism and to set the timeout to a very low value:

  static int
  snmp_setup_session_cb(int majorID, int minorID,
                void *serverarg, void *clientarg)
   
      netsnmp_session *sess = serverarg;
      sess->timeout = ONE_SEC / 3;
      sess->retries = 0;
      return 0;
   
  void some_init_function()
   
      /* [...] */
      snmp_register_callback(SNMP_CALLBACK_LIBRARY,
         SNMP_CALLBACK_SESSION_INIT,
         snmp_setup_session_cb, NULL);
      /* [...] */
   
  

Unresponsive master agent Why is the master agent unresponsive? snmpd is also an event-based program and, trust me, there are a lot of places where it will become unresponsive. Here are two examples:

• When using the pass or the pass_persist directive, Net-SNMP will block until the delegated command outputs a complete line.

  /* var_extensible_pass() in agent/mibgroup/ucd-snmp/pass.c */
  /*
   * valid call.  Exec and get output 
   */
  if ((fd = get_exec_output(passthru)) != -1)  
      file = fdopen(fd, "r");
      if (fgets(buf, sizeof(buf), file) == NULL)  
          fclose(file);
          wait_on_exec(passthru);
          /* [...] */
          continue;
       
      /* [...] */
      fclose(file);
      wait_on_exec(passthru);
      /* [...] */
   
  

• DISMAN-PING-MIB implementation will just hang while pinging the remote host. If the host is down and you have requested 5 probes with a timeout of 1 second, snmpd will be unresponsive for 5 seconds. A similar problem exists with DISMAN-TRACEROUTE-MIB.

Minimal master agent More and more code is running in the main agent. You cannot expect it to not block. But we could add a minimal master agent to the mix: it will handle only a few MIB modules by itself while most MIB modules will be delagated to a complete agent running as a subagent. Your own subagent will connect to the minimal master agent. It is quite easy to turn snmpd into a subagent: just use -X flag. We should also disable unwanted MIB modules for each agent with -I flag. For the minimal master agent, we only enable some modules. All other modules³ will be served by snmpd acting as a subagent.

$ MODS="snmp_mib,sysORTable,usmConf,usmStats,usmUser,vacm_conf,vacm_context,vacm_vars"
$ snmpd -Lsd -Lf /dev/null -u snmp -g snmp \
>   -C -c /etc/snmp/snmpd.master.conf \
>   -p /var/run/snmpd.master.pid \
>   -I $MODS
$ snmpd -Lsd -Lf /dev/null -u snmp -g snmp \
>   -p /var/run/snmpd.pid -X \
>   -I -$MODS

Access control configuration should be placed in /etc/snmp/snmpd.master.conf, as well as master agentx directive. Other directives (like pass, pass_persist, load, sysname, ) in /etc/snmp/snmpd.conf. The minimal master agent is very unlikely to block and while the ground cause is still here, it is now very much mitigated. You may extend this solution by splicing snmpd into several subagents: one handling safe modules, one handling pass and pass_persist stuff and a last one for unsafe modules (like DISMAN-PING-MIB).

---

1. agentx_reopen_session() will call agentx_close_session() and agentx_open_session(). agentx_register() and the following functions will be called by those two functions.
2. NETSNMP_DS_AGENT_AGENTX_TIMEOUT and NETSNMP_DS_AGENT_AGENTX_RETRIES variables are only used by the master agent, not by a subagent.
3. Use snmpd -Dmib_init -H to get the list of modules. You may want to add smux module to the master agent.

What is it? Last month I started what I am calling the LUV Hardware Library. It s a briefcase full of computer parts that are free to LUV members which I plan to bring to all LUV meetings [1]. The issue is that there is a lot of hardware which has no great value to people and is often excess to requirements but which is still somewhat expensive or difficult to obtain and thus transferring it from the people who don t need it to people who do provides a significant benefit to the recipient but no real cost to the donor. Currently my briefcase has a range of different types of DIMM, some PCI cards, some assorted cables, a laptop SATA disk, and a bunch of other random things. Most of the stuff has been in my spare parts pile for a year or two, but some of it has been donated by other people. What we Need The next thing we need is more packaging, anti-static bags for RAM and PCI cards, sealable bags for screws, and some way of storing motherboard batteries. In terms of hardware that can be donated one thing that would be really useful is SATA disks. I m sure that there are lots of people who have upgraded the storage of their computer and kept the old drive with no real chance of using it. But there are lots of people who need such disks, for example I ve been giving away a bunch of P4 systems which lack disks and are not cabled for IDE disks a supply of SATA disks would make them a lot more usable. Also any other random electronic stuff is of interest, including non-Linux things such as mobile phones (but only if not network locked). If you have something that s big or heavy to donate then contact me via email first. Other Options Computerbank does great work in rebuilding old PCs and selling them cheaply to worthy people [2], but most of the spare hardware I get is below the minimum specs that they will accept. I m not planning to compete with Computerbank in any way, I just want to provide a useful service to LUV members who want to upgrade their PC for free. I encourage other people to do the same at other LUG meetings!

• [1] http://luv.asn.au/
• [2] http://www.computerbank.org.au/

Related posts:

1. Donating old Hardware On a recent visit to my local e-waste disposal place...
2. Giving Away Hardware For the last few years I have been actively seeking...
3. Shelf-life of Hardware Recently I ve been having some problems with hardware dying. Having...

During weekend, I had great opportunity not only to visit Prague Airport, but also to taste airspotting. As photography is my hobby, this was interesting experience, though I don't think it would be my favorite thing to capture. Anyway we could see some interesting places - L1011 TriStar KC1 from Royal Air Force, two Boeings 747-400F and Boeing 777-300. The Boeings being the biggest airplanes you can regularly see in Prague. So lets stop talking and here are few pictures:

Filed under: English Photography 0 comments Flattr this!

In mailing list discussions, I've noticed that the practice of replying "+1" to messages with which one agrees has become very widespread. I think this is interesting for a lot of reasons, such as what it shows about habits picked up in communicating via social network sites that grew out of e-mail percolating back to e-mail. We seem to have really grown to like this idea of "voting" for people's messages, and if there's no "Like" or "+1" button on their message, we'll make our own. (Likewise, apparently a person's name is no longer sufficient to indicate that one is addressing them in an e-mail -- am I right, @dear_reader?) The problem for me is that I have my venerable awesome e-mail client Gnus set to hide quoted text in messages by default with (add-hook 'gnus-article-prepare-hook (lambda () (gnus-article-hide-citation 1))), and it thinks that "+" is the start of a quoted line -- so it hides the "+1". While there's a certain poetic justice to that, it confuses me because it makes the message look empty, and someone could do something like "+1 I'm pregnant too", causing me to miss some very important news. The fix for this is to change the regexp pattern Gnus uses to decide if a line is a message quote or not. gnus-message-cite-prefix-regexp is built using message-cite-prefix-regexp, so I changed the value of the latter, removing the literal "+". I also found while investigating this that I already configure that variable in order to add "#>" which someone sometime during my 9 years of using Gnus must have used. The net result is (setq message-cite-prefix-regexp "\\([ ]*[-_.#[:word:]]+>+\\ [ ]*[]> ]\\)+"). (Note that some of those whitespace characters are tabs, which probably won't display properly here, but you can compare to the default value to see what I actually changed). +1?