OpenSSH has this very nice setting, VerifyHostKeyDNS, which when enabled, will pull SSH host keys from DNS, and you no longer need to either trust on first use, or copy host keys around out of band. Naturally, trusting unsecured DNS is a bit scary, so this requires the record to be signed using DNSSEC. This has worked for a long time, but then broke, seemingly out of the blue. Running ssh -vvv gave output similar to

debug1: found 4 insecure fingerprints in DNS
debug3: verify_host_key_dns: checking SSHFP type 1 fptype 2
debug3: verify_host_key_dns: checking SSHFP type 4 fptype 2
debug1: verify_host_key_dns: matched SSHFP type 4 fptype 2
debug3: verify_host_key_dns: checking SSHFP type 4 fptype 1
debug1: verify_host_key_dns: matched SSHFP type 4 fptype 1
debug3: verify_host_key_dns: checking SSHFP type 1 fptype 1
debug1: matching host key fingerprint found in DNS

even though the zone was signed, the resolver was checking the signature and I even checked that the DNS response had the AD bit set. The fix was to add options trust-ad to /etc/resolv.conf. Without this, glibc will discard the AD bit from any upstream DNS servers. Note that you should only add this if you actually have a trusted DNS resolver. I run unbound on localhost, so if somebody can do a man-in-the-middle attack on that traffic, I have other problems.

I moved my blog around a bit and it appears that static pages are now in favour, so I switched to that, by way of Hugo. CSS and such needs more tweaking, but it ll make do for now. As part of this, RSS feeds and such changed, if you want to subscribe to this (very seldomly updated) blog, use https://err.no/personal/blog/index.xml

What happened in the reproducible builds effort this week: Toolchain fixes

• Markus Koschany uploaded antlr3/3.5.2-3 which includes a fix by Emmanuel Bourg to make the generated parser reproducible.
• Markus Koschany uploaded maven-bundle-plugin/2.4.0-2 which includes a fix by Emmanuel Bourg to use the date in the DEB_CHANGELOG_DATETIME variable in the pom.properties file embedded in the jar files.
• Niels Thykier uploaded debhelper/9.20151116 which makes the timestamp of directories created by dh_install, dh_installdocs, and dh_installexamples reproducible. Patch by Niko Tyni.

Mattia Rizzolo uploaded a version of perl to the reproducible repository including the patch written by Niko Tyni to add support for SOURCE_DATE_EPOCH in Pod::Man. Dhole sent an updated version of his patch adding support for SOURCE_DATE_EPOCH in GCC to the upstream mailing list. Several comments have been made in response which have been quickly addressed by Dhole. Dhole also forwarded his patch adding support for SOURCE_DATE_EPOCH in libxslt upstream. Packages fixed The following packages have become reproducible due to changes in their build dependencies: antlr3/3.5.2-3, clusterssh, cme, libdatetime-set-perl, libgraphviz-perl, liblingua-translit-perl, libparse-cpan-packages-perl, libsgmls-perl, license-reconcile, maven-bundle-plugin/2.4.0-2, siggen, stunnel4, systemd, x11proto-kb. The following packages became reproducible after getting fixed:

• bindex/2.2+svn101-3 by Markus Koschany.
• glyr/1.0.8-2 by Etienne Millon.
• jenkins-json/2.4-jenkins-3-4 by Emmanuel Bourg.
• pkg-config/0.29-1 uploaded by Tollef Fog Heen, original patch by Juan Picca.
• plexus-containers1.5/1.6-1 by Emmanuel Bourg.
• polyglot-maven/0.8~tobrien+git20120905-5 by Emmanuel Bourg.
• sigil/0.9.0+dfsg-3 uploaded by Mattia Rizzolo, original patch by Reiner Herrmann.
• simutrans/120.1.1+repack-2 uploaded by J rg Frings-F rst, fix by Markus Koschany.
• torrus/2.08-4 by Bernhard Schmidt.
• trigger-rally-data/0.6.1-2 uploaded by Bertrand Marc, patch by Mattia Rizzolo.

Some uploads fixed some reproducibility issues, but not all of them:

• castle-game-engine/5.2.0-1 by Paul Gevers.
• libam7xxx/0.1.6-2 by Antonio Ospite.
• libpdfbox-java/1:1.8.10-1 by Emmanuel Bourg.
• xfaces/3.3-29 uploaded by Hakan Ardo, original patch by Chris Lamb.

reproducible.debian.net Vagrant Cascadian has set up a new armhf node using a Raspberry Pi 2. It should soon be added to the Jenkins infrastructure. diffoscope development diffoscope version 42 was release on November 20th. It adds a missing dependency on python3-pkg-resources and to prevent similar regression another autopkgtest to ensure that the command line is functional when Recommends are not installed. Two more encoding related problems have been fixed (#804061, #805418). A missing Build-Depends has also been added on binutils-multiarch to make the test suite pass on architectures other than amd64. Package reviews 180 reviews have been removed, 268 added and 59 updated this week. 70 new fail to build from source bugs have been reported by Chris West, Chris Lamb and Niko Tyni. New issue this week: randomness_in_ocaml_preprocessed_files. Misc. Jim MacArthur started to work on a system to rebuild and compare packages built on reproducible.debian.net using .buildinfo and snapshot.debian.org. On December 1-3rd 2015, a meeting of about 40 participants from 18 different free software projects will be held in Athens, Greece with the intent of improving the collaboration between projects, helping new efforts to be started, and brainstorming on end-user aspects of reproducible builds.

I've had a half-broken temperature monitoring setup at home for quite some time. It started out with a Atom-based NAS, a USB-serial adapter and a passive 1-wire adapter. It sometimes worked, then stopped working, then started when poked with a stick. Later, the NAS was moved under the stairs and I put a Beaglebone Black in its old place. The temperature monitoring thereafter never really worked, but I didn't have the time to fix it. Over the last few days, I've managed to get it working again, of course by replacing nearly all the existing components. I'm using the DS18B20 sensors. They're about USD 1 a piece on Ebay (when buying small quantities) and seems to work quite ok. My first task was to address the reliability problems: Dropouts and really poor performance. I thought the passive adapter was problematic, in particular with the wire lengths I'm using and I therefore wanted to replace it with something else. The BBB has GPIO support, and various blog posts suggested using that. However, I'm running Debian on my BBB which doesn't have support for DTB overrides, so I needed to patch the kernel DTB. (Apparently, DTB overrides are landing upstream, but obviously not in time for Jessie.) I've never even looked at Device Tree before, but the structure was reasonably simple and with a sample override from bonebrews it was easy enough to come up with my patch. This uses pin 11 (yes, 11, not 13, read the bonebrews article for explanation on the numbering) on the P8 block. This needs to be compiled into a .dtb. I found the easiest way was just to drop the patched .dts into an unpacked kernel tree and then running make dtbs. Once this works, you need to compile the w1-gpio kernel module, since Debian hasn't yet enabled that. Run make menuconfig, find it under "Device drivers", "1-wire", "1-wire bus master", build it as a module. I then had to build a full kernel to get the symversions right, then build the modules. I think there is or should be an easier way to do that, but as I cross-built it on a fast AMD64 machine, I didn't investigate too much. Insmod-ing w1-gpio then works, but for me, it failed to detect any sensors. Reading the data sheet, it looked like a pull-up resistor on the data line was needed. I had enabled the internal pull-up, but apparently that wasn't enough, so I added a 4.7kOhm resistor between pin 3 (VDD_3V3) on P9 and pin (GPIO_45) on P8. With that in place, my sensors showed up in /sys/bus/w1/devices and you can read the values using cat. In my case, I wanted the data to go into collectd and then to graphite. I first tried using an Exec plugin, but never got it to work properly. Using a [python plugin] worked much better and my graphite installation is now showing me temperatures. Now I just need to add more probes around the house. The most useful references were

• Bonebrews article
• Cross compiling official kernel packages
• Verifying pin usage. This is useful for double checking that your DTB actually exports the pins you expect it to.

In addition, various searches for DS18B20 pinout and similar, of course.

Apparently, people care when you, as privileged person (white, male, long-time Debian Developer) throw in the towel because the amount of crap thrown your way just becomes too much. I guess that's good, both because it gives me a soap box for a short while, but also because if enough people talk about how poisonous the well that Debian is has become, we can fix it. This morning, I resigned as a member of the systemd maintainer team. I then proceeded to leave the relevant IRC channels and announced this on twitter. The responses I've gotten have been almost all been heartwarming. People have generally been offering hugs, saying thanks for the work put into systemd in Debian and so on. I've greatly appreciated those (and I've been getting those before I resigned too, so this isn't just a response to that). I feel bad about leaving the rest of the team, they're a great bunch: competent, caring, funny, wonderful people. On the other hand, at some point I had to draw a line and say "no further". Debian and its various maintainer teams are a bunch of tribes (with possibly Debian itself being a supertribe). Unlike many other situations, you can be part of multiple tribes. I'm still a member of the DSA tribe for instance. Leaving pkg-systemd means leaving one of my tribes. That hurts. It hurts even more because it feels like a forced exit rather than because I've lost interest or been distracted by other shiny things for long enough that you don't really feel like part of a tribe. That happened with me with debian-installer. It was my baby for a while (with a then quite small team), then a bunch of real life thing interfered and other people picked it up and ran with it and made it greater and more fantastic than before. I kinda lost touch, and while it's still dear to me, I no longer identify as part of the debian-boot tribe. Now, how did I, standing stout and tall, get forced out of my tribe? I've been a DD for almost 14 years, I should be able to weather any storm, shouldn't I? It turns out that no, the mountain does get worn down by the rain. It's not a single hurtful comment here and there. There's a constant drum about this all being some sort of conspiracy and there are sometimes flares where people wish people involved in systemd would be run over by a bus or just accusations of incompetence. Our code of conduct says, "assume good faith". If you ever find yourself not doing that, step back, breathe. See if there's a reasonable explanation for why somebody is saying something or behaving in a way that doesn't make sense to you. It might be as simple as your native tongue being English and their being something else. If you do genuinely disagree with somebody (something which is entirely fine), try not to escalate, even if the stakes are high. Examples from the last year include talking about this as a war and talking about "increasingly bitter rear-guard battles". By using and accepting this terminology, we, as a project, poison ourselves. Sam Hartman puts this better than me:

I'm hoping that we can all take a few minutes to gain empathy for those who disagree with us. Then I'm hoping we can use that understanding to reassure them that they are valued and respected and their concerns considered even when we end up strongly disagreeing with them or valuing different things.

I'd be lying if I said I didn't ever feel the urge to demonise my opponents in discussions. That they're worse, as people, than I am. However, it is imperative to never give in to this, since doing that will diminish us as humans and make the entire project poorer. Civil disagreements with reasonable discussions lead to better technical outcomes, happier humans and a healthier projects.

The GNOME and many other infrastructures have been recently attacked by an huge amount of subscription-based spam against their Mailman istances. What the attackers were doing was simply launching a GET call against a specific REST API URL passing all the parameters it needed for a subscription request (and confirmation) to be sent out. Understanding it becomes very easy when you look at the following example taken from our apache.log:

May 3 04:14:38 restaurant apache: 81.17.17.90, 127.0.0.1 - - [03/May/2014:04:14:38 +0000] "GET /mailman/subscribe/banshee-list?email=example@me.com&fullname=&pw=123456789&pw-conf=123456789&language=en&digest=0&email-button=Subscribe HTTP/1.1" 403 313 "http://spam/index2.html" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/34.0.1847.131 Safari/537.36"

As you can the see attackers were sending all the relevant details needed for the subscription to go forward (and specifically the full name, the email, the digest option and the password for the target list). At first we tried to either stop the spam by banning the subnets where the requests were coming from, then when it was obvious that more subnets were being used and manual intervention was needed we tried banning their User-Agents. Again no luck, the spammers were smart enough to change it every now and then making it to match an existing browser User-Agent. (with a good percentage to have a lot of false-positives) Now you might be wondering why such an attack caused a lot of issues and pain, well, the attackers made use of addresses found around the web for their malicius subscription requests. That means we received a lot of emails from people that have never heard about the GNOME mailing lists but received around 10k subscription requests that were seemingly being sent by themselves. It was obvious we needed to look at a backup solution and luckily someone on our support channel suggested the freedesktop.org sysadmins recently added CAPTCHAs support to Mailman. I m now sharing the patch and providing a few more details on how to properly set it up on either DEB or RPM based distributions. Credits for the patch should be given to Debian Developer Tollef Fog Heen, who has been so kind to share it with us. Before patching your installation make sure to install the python-recaptcha package (tested on Debian with Mailman 2.1.15) on DEB based distributions and python-recaptcha-client on RPM based distributions. (I personally tested it against Mailman release 2.1.15, RHEL 6) The Patch

diff --git a/Mailman/Cgi/listinfo.py b/Mailman/Cgi/listinfo.py
index 4a54517..d6417ca 100644
--- a/Mailman/Cgi/listinfo.py
+++ b/Mailman/Cgi/listinfo.py
@@ -22,6 +22,7 @@
 
 import os
 import cgi
+import sys
 
 from Mailman import mm_cfg
 from Mailman import Utils
@@ -30,6 +31,8 @@ from Mailman import Errors
 from Mailman import i18n
 from Mailman.htmlformat import *
 from Mailman.Logging.Syslog import syslog
+sys.path.append("/usr/share/pyshared")
+from recaptcha.client import captcha
 
 # Set up i18n
 _ = i18n._
@@ -200,6 +203,9 @@ def list_listinfo(mlist, lang):
     replacements[''] = mlist.FormatFormStart('listinfo')
     replacements[''] = mlist.FormatBox('fullname', size=30)
 
+    # Captcha
+    replacements[''] = captcha.displayhtml(mm_cfg.RECAPTCHA_PUBLIC_KEY, use_ssl=False)
+
     # Do the expansion.
     doc.AddItem(mlist.ParseTags('listinfo.html', replacements, lang))
     print doc.Format()
diff --git a/Mailman/Cgi/subscribe.py b/Mailman/Cgi/subscribe.py
index 7b0b0e4..c1c7b8c 100644
--- a/Mailman/Cgi/subscribe.py
+++ b/Mailman/Cgi/subscribe.py
@@ -21,6 +21,8 @@ import sys
 import os
 import cgi
 import signal
+sys.path.append("/usr/share/pyshared")
+from recaptcha.client import captcha
 
 from Mailman import mm_cfg
 from Mailman import Utils
@@ -132,6 +130,17 @@ def process_form(mlist, doc, cgidata, lang):
     remote = os.environ.get('REMOTE_HOST',
                             os.environ.get('REMOTE_ADDR',
                                            'unidentified origin'))
+
+    # recaptcha
+    captcha_response = captcha.submit(
+        cgidata.getvalue('recaptcha_challenge_field', ""),
+        cgidata.getvalue('recaptcha_response_field', ""),
+        mm_cfg.RECPTCHA_PRIVATE_KEY,
+        remote,
+        )
+    if not captcha_response.is_valid:
+        results.append(_('Invalid captcha'))
+
     # Was an attempt made to subscribe the list to itself?
     if email == mlist.GetListEmail():
         syslog('mischief', 'Attempt to self subscribe %s: %s', email, remote)

Additional setup Then on the /var/lib/mailman/templates/en/listinfo.html template (right below <mm-digest-question-end>) add:

      <tr>
        <td>Please fill out the following captcha</td>
        <td><mm-recaptcha-javascript></TD>
      </tr>

Make also sure to generate a public and private key at https://www.google.com/recaptcha and add the following paramaters on your mm_cfg.py file:

• RECAPTCHA_PRIVATE_KEY
• RECAPTCHA_PUBLIC_KEY

Loading reCAPTCHAs images from a trusted HTTPS source can be done by changing the following line:

replacements[''] = captcha.displayhtml(mm_cfg.RECAPTCHA_PUBLIC_KEY, use_ssl=False)

to

replacements[''] = captcha.displayhtml(mm_cfg.RECAPTCHA_PUBLIC_KEY, use_ssl=True)

EPEL 6 related details A few additional details should be provided in case you are setting this up against a RHEL 6 host: (or any other machine using the EPEL 6 package python-recaptcha-client-1.0.5-3.1.el6) Importing the recaptcha.client module will fail for some strange reason, importing it correctly can be done this way:

ln -s /usr/lib/python2.6/site-packages/recaptcha/client /usr/lib/mailman/pythonlib/recaptcha

and then fix the imports also making sure sys.path.append( /usr/share/pyshared ) is not there:

from recaptcha import captcha

That s not all, the package still won t work as expected given the API_SSL_SERVER, API_SERVER and VERIFY_SERVER variables on captcha.py are outdated (filed as bug #1093855), substitute them with the following ones:

API_SSL_SERVER="https://www.google.com/recaptcha/api"
API_SERVER="http://www.google.com/recaptcha/api"
VERIFY_SERVER="www.google.com"

And then on line 76:

url = "https://%s/recaptcha/api/verify" % VERIFY_SERVER,

That should be all! Enjoy!

Dustin Kirkland wrote an interesting post about fingerprint authentication [1]. He suggests using fingerprints for identifying users (NOT authentication) and gives an example of a married couple sharing a tablet and using fingerprints to determine who s apps are loaded. In response Tollef Fog Heen suggests using fingerprints for lightweight authentication, such as resuming a session after a toilet break [2]. I think that one of the best comments on the issue of authentication for different tasks is in XKCD comic 1200 [3]. It seems obvious that the division between administrator (who installs new device drivers etc) and user (who does everything from playing games to online banking with the same privileges) isn t working, and never could work well particularly when the user in question installs their own software. I think that one thing which is worth considering is the uses of a signature. A signature can be easily forged in many ways and they often aren t checked well. It seems that there are two broad cases of using a signature, one is to enter into legally binding serious contract such as a mortgage (where wanting to sign is the relevant issue) and the other is cases where the issue doesn t matter so much (EG signing off on a credit card purchase where the parties at risk can afford to lose money on occasion for efficient transactions). Signing is relatively easy but that s because it either doesn t matter much or because it s just a legal issue which isn t connected to authentication. The possibility of serious damage (sending life savings or incriminating pictures to criminals in another jurisdiction) being done instantly never applied to signatures. It seems to me that in many ways signatures are comparable to fingerprints and both of them aren t particularly good for authentication to a computer. In regard to Tollef s ideas about lightweight authentication I think that the first thing that would be required is direct user control over the authentication required to unlock a system. I have read about some Microsoft research into a computer monitoring the office environment to better facilitate the user s requests, an obvious extension to such research would be to have greater unlock requirements if there are more unknown people in the area or if the device is in a known unsafe location. But apart from that sort of future development it seems that having the user request a greater or lesser authentication check either at the time they lock their session or by policy would make sense. Generally users have a reasonable idea about the risk of another user trying to login with their terminal so user should be able to decide that a toilet break when at home only requires a fingerprint (enough to keep out other family members) while a toilet break at the office requires greater authentication. Mobile devices could use GPS location to determine unlock requirements, GPS can be forged, but if your attacker is willing and able to do that then you have a greater risk than most users. Some users turn off authentication on their phone because it s too inconvenient. If they had the option of using a fingerprint most of the time and a password for the times when a fingerprint can t be read then it would give an overall increase in security. Finally it should be possible to unlock only certain applications. Recent versions of Android support widgets on the lock screen so you can perform basic tasks such as checking the weather forecast without unlocking your phone. But it should be possible to have different authentication requirements for various applications. Using a fingerprint scan to allow playing games or reading email in the mailing list folder would be more than adequate security. But reading the important email and using SMS probably needs greater authentication. This takes us back to the XKCD cartoon.

• [1] http://blog.dustinkirkland.com/2013/10/fingerprints-are-user-names-not.html
• [2] http://tinyurl.com/lak528z
• [3] http://xkcd.com/1200/

I'm running a local unbound instance on my laptop to get working DNSSEC. It turns out that with the captive portal NSB (the Norwegian national rail company), this doesn't work too well and you get into an endless series of redirects. Changing resolv.conf so you use the DHCP-provided resolver stops the redirect loop and you can then log in. Afterwards, you're free to switch back to using your own local resolver.

Dustin Kirkland recently wrote that "Fingerprints are usernames, not passwords". I don't really agree, I think fingerprints are fine for lightweight authentication. iOS at least allows you to only require a pass code after a time period has expired, so you don't have to authenticate to the phone all the time. Replacing no authentication with weak authentication (but only for a fairly short period) will improve security over the current status, even if it's not perfect. Having something similar for Linux would also be reasonable, I think. Allow authentication with a fingerprint if I've only been gone for lunch (or maybe just for a trip to the loo), but require password or token if I've been gone for longer. There's a balance to be struck between convenience and security.

NSCA is a tool used to submit passive check results to nagios. Unfortunately, an incompatibility was recently introduced between wheezy clients and old servers. Since I don't want to upgrade my server, this caused some problems and I decided to just get rid of NSCA completely. The server side of NSCA is pretty trivial, it basically just adds a timestamp and a command name to the data sent by the client, then changes tabs into semicolons and stuffs all of that down Nagios' command pipe. The script I came up with was:

#! /usr/bin/python
# -* coding: utf-8 -*-
import time
import sys
# format is:
# [TIMESTAMP] COMMAND_NAME;argument1;argument2; ;argumentN
#
# For passive checks, we want PROCESS_SERVICE_CHECK_RESULT with the
# format:
#
# PROCESS_SERVICE_CHECK_RESULT;<host_name>;<service_description>;<return_code>;<plugin_output>
#
# return code is 0=OK, 1=WARNING, 2=CRITICAL, 3=UNKNOWN
#
# Read lines from stdin with the format:
# $HOSTNAME\t$SERVICE_NAME\t$RETURN_CODE\t$TEXT_OUTPUT
if len(sys.argv) != 2:
    print "Usage:  0  HOSTNAME".format(sys.argv[0])
    sys.exit(1)
HOSTNAME = sys.argv[1]
timestamp = int(time.time())
nagios_cmd = file("/var/lib/nagios3/rw/nagios.cmd", "w")
for line in sys.stdin:
    (_, service, return_code, text) = line.split("\t", 3)
    nagios_cmd.write(u"[ timestamp ] PROCESS_SERVICE_CHECK_RESULT; hostname ; service ; return_code ; text \n".format
                     (timestamp = timestamp,
                      hostname = HOSTNAME,
                      service = service,
                      return_code = return_code,
                      text = text))

The reason for the hostname in the line (even though it's overridden) is to be compatible with send_nsca's input format. Machines submit check results over SSH using its excellent ForceCommand capabilities, the Chef template for the authorized_keys file looks like:

<% for host in @nodes %>
command="/usr/local/lib/nagios/nagios-passive-check-result <%= host[:hostname] %>",no-port-forwarding,no-X11-forwarding,no-agent-forwarding,no-pty ssh-rsa <%= host[:keys][:ssh][:host_rsa_public] %> <%= host[:hostname] %>
<% end %>

The actual chef recipe looks like:

nodes = []
search(:node, "*:*") do  n 
  # Ignore not-yet-configured nodes                                                                       
  next unless n[:hostname]
  next unless n[:nagios]
  next if n[:nagios].has_key?(:ignore)
  nodes << n
end
nodes.sort!    a,b  a[:hostname] <=> b[:hostname]  
print nodes
template "/etc/ssh/userkeys/nagios" do
  source "authorized_keys.erb"
  mode 0400
  variables( 
              :nodes => nodes
             )
end
cookbook_file "/usr/local/lib/nagios/nagios-passive-check-result" do
  mode 0555
end
user "nagios" do
  action :manage
  shell "/bin/sh"
end

To submit a check, hosts do:

printf "$HOSTNAME\t$SERVICE_NAME\t$RET\t$TEXT\n"   ssh -i /etc/ssh/ssh_host_rsa_key -o BatchMode=yes -o StrictHostKeyChecking=no -T nagios@$NAGIOS_SERVER

Recently, there's been discussions on IRC and the debian-devel mailing list about how to notify users, typically from a cron script or a system daemon needing to tell the user their hard drive is about to expire. The current way is generally "send email to root" and for some bits "pop up a notification bubble, hoping the user will see it". Emailing me means I get far too many notifications. They're often not actionable (apt-get update failed two days ago) and they're not aggregated. I think we need a system that at its core has level and edge triggers and some way of doing flap detection. Level interrupts means "tell me if a disk is full right now". Edge means "tell me if the checksums have changed, even if they now look ok". Flap detection means "tell me if the nightly apt-get update fails more often than once a week". It would be useful if it could extrapolate some notifications too, so it could tell me "your disk is going to be full in $period unless you add more space". The system needs to be able to take in input in a variety of formats: syslog, unstructured output from cron scripts (including their exit codes), snmp, nagios notifications, sockets and fifos and so on. Based on those inputs and any correlations it can pull out of it, it should try to reason about what's happening on the system. If the conclusion there is "something is broken", it should see if it's something that it can reasonably fix by itself. If so, fix it and record it (so it can be used for notification if appropriate: I want to be told if you restart apache every two minutes). If it can't fix it, notify the admin. It should also group similar messages so a single important message doesn't drown in a million unimportant ones. Ideally, this should be cross-host aggregation. The notifications should be possible to escalate if they're not handled within some time period. I'm not aware of such a tool. Maybe one could be rigged together by careful application of logstash, nagios, munin/ganglia/something and sentry. If anybody knows of such a tool, let me know, or if you're working on one, also please let me know.

Hello everyone! I am very excited to report about the awesome progress we made with Tanglu, the new Debian-based Linux-Distribution. First of all, some off-topic info: I don t feel comfortable with posting too much Tanglu stuff to Planet-KDE, as this is often not KDE-related. So, in future Planet-KDE won t get Tanglu information, unless it is KDE-related You might want to take a look at Planet-Tanglu for (much) more information. So, what happened during the last weeks? Because I haven t had lectures, I nearly worked full-time on Tanglu, setting up most of the infrastructure we need. (this will change with the next week, where I have lectures again, and I also have work to do on other projects, not just Tanglu ^^) Also, we already have an awesome community of translators, designers and developers. Thanks to them, the Tanglu-Website is now translated to 6 languages, more are in the pipeline and will be merged later. Also, a new website based on the Django framework is in progress. The Logo-Contest We ve run a logo-contest, to find a new and official Tanglu logo, as the previous logo draft was too close to the Debian logo (I asked the trademark people at Debian). More than 30 valid votes (you had to be subscribed to a Tanglu Mailinglist) were received for 7 logo proposals, and we now have a final logo: I like it very much Fun with dak I decided to use dak, the Debian Archive Kit, to handle the Tanglu archive. Choosing dak over smaller and easy-to-use solutions had multiple reasons, the main reason is that dak is way more flexible than the smaller solution (like reprepro or min-dak) and able to handle the large archive of Tanglu. Also, dak is lightning fast. And I would have been involved with dak sooner or later anyway, because I will implement the DEP-11 extension to the Debian Archive later (making the archive application-friendly). Working with dak is not exactly fun. The documentation is not that awesome, and dak contains many hardcoded stuff for Debian, e.g. it often expects the unstable suite to be present. Also, running dak on Debian Wheezy turned out to be a problem, as the Python module apt_pkg changed the API and dak naturally had problems with that. But with the great help of some Debian ftpmasters (many thanks to that!), dak is now working for Tanglu, managing the whole archive. There are still some quirks which need to be fixed, but the archive is in an usable state, accepting and integrating packages. The work on dak is also great for Debian: I resolved many issues with non-Debian dak installations, and made many parts of dak Wheezy-proof. Also, I added a few functions which might also be useful for Debian itself. All patches have of course been submitted to upstream-dak. Wanna-build and buildds This is also nearly finished Wanna-build, the software which manages all buildds for an archive, is a bit complicated to use. I still have some issues with it, but it does it s job so far. (need to talk to the Debian wanna-build admins for help, e.g. wanna-build seems to be unable to handle arch:all-only packages, also, build logs are only submitted in parts) The status of Tanglu builds can be viewed at the provisoric Buildd-Status pages. Setting up a working buildd is also a tricky thing, it involves patching sbuild to escape bug #696841 and applying various workarounds to make the buildd work and upload packages correctly. I will write instructions how to set-up and maintain a buildd soon. At time, we have only one i386 buildd up and running, but more servers (in numbers: 3) are prepared and need to be turned into buildds. After working on Wanna-build and dak, I fully understand why Canonical developed Launchpad and it s Soyuz module for Ubuntu. But I think we might be able to achieve something similar, using just the tools Debian already uses (maybe a little less confortable than LP, but setting up an own LP instance would have been much more trouble). Debian archive import The import of packages from the Debian archive has finished. Importing the archive resulted in many issues and some odd findings (I didn t know that there are packages in the archive which didn t receive an upload since 2004!), but it has finally finished, and the archive is in a consistent state at time. To have a continuous package import from Debian while a distribution is in development, we need some changes to wanna-build, which will hopefully be possible. Online package search The Online-Package-Search is (after resolving many issues, who expected that? ) up and running. You can search for any package there. Some issues are remaining, e.g. the file-contents-listing doesn t work, and changelog support is broken, but the basic functionality is there. Tanglu Bugtracker We now also have a bugtracker which is based on the Trac software. The Tanglu-Bugtracker is automatically synced with the Tanglu archive, meaning that you find all packages in Trac to report bugs against them. The dak will automatically update new packages every day. Trac still needs a few confort-adjustments, e.g. submitting replies via email or tracking package versions. Tanglu base system The Tanglu metapackages have been published in a first alpha version. We will support GNOME-3 and KDE4, as long as this is possible (= enough people working on the packaging). The Tanglu packages will also depend on systemd, which we will need in GNOME anyway, and which also allows some great new features in KDE. Side-effect of using systemd is at least for the start that Tanglu boots a bit slow, because we haven t done any systemd adjustments, and because systemd is very old. We will have to wait for the systemd and udev maintainers to merge the packages and release a new version first, before this will improve. (I don t want to do this downstream in Tanglu, because I don t know the plans for that at Debian (I only know the information Tollef Fog Heen & Co. provided at FOSDEM)) The community The community really surprised me! We got an incredible amount of great feedback on Tanglu, and most of the people liked the idea of Tanglu. I think we are one of the less-flamed new distributions ever started . Also, without the very active community, kickstarting Tanglu would not have been possible. My guess was that we might be able to have something running next year. Now, with the community help, I see a chance for a release in October The only thing people complained about was the name of the distribution. And to be really honest, I am not too happy with the name. But finding a name was an incredibe difficult process (finding something all parties liked), and Tanglu was a good compromise. Tanglu has absolutely no meaning, it was taken because it sounded interesting. The name was created by combining the Brazilian Tangerine (Clementine) and the German Iglu (igloo). I also dont think the name matters that much, and I am more interested in the system itself than the name of it. Also, companies produce a lot of incredibly weird names, Tanglu is relatively harmless compared to that . In general, thanks to everyone participating in Tanglu! You are driving the project forward! The first (planned) release I hereby announce the name of the first Tanglu release, 1.1 Aequorea Victoria . It is Daniel s fault that Tanglu releases will be named after jellyfishes, you can ask him why if you want I picked Aequorea, because this kind of jellyfish was particularly important for research in molecular biology. The GFP protein, a green fluorescent protein (=GFP), caused a small revolution in science and resulted in a Nobel Prize in 2008 for the researchers involved in GFP research (for the interested people: You can tag proteins with GFP and determine their position using light microscopy. GFP also made many other fancy new lab methods possible). Because Tanglu itself is more or less experimental at time, I found the connection to research just right for the very first release. We don t have a time yet when this version will be officially released, but I expect it to be in October, if the development speed increases a little and more developers get interested and work on it. Project Policy We will need to formalize the Tanglu project policy soon, both the technical and the social policies. In general, regarding free software and technical aspects, we strictly adhere to the Dbian Free Software Guidelines, the Debian Social Contract and the Debian Policy. Some extra stuff will be written later, please be patient! Tanglu OIN membership I was approached by the Open Invention Network, to join it as member. In general, I don t have objections to do that, because it will benefit Tanglu. However, the OIN has a very tolerant public stance on software patents, which I don t like that much. Debian did not join the OIN for this reason. For Tanglu, I think we could still join the OIN without someone thinking that we support the stance on software patents. Joining would simply be pragmatic: We support the OIN as a way to protect the Linux ecosystem from software patents, even if we don t like the stance on software patents and see it differently. Because this affects the whole Tanglu project, I don t want to decide this alone, but get some feedback from the Tanglu community before making a decision. Can I install Tanglu now? Yes and no. We don t provide installation images yet, so trying Tanglu is a difficult task (you need to install Debian and then upgrade it to Tanglu) if you want to experiment with it, I recomment trying Tanglu in a VM. I want to help! Great, then please catch one of us on IRC or subscribe to the mailinglists. The best thing is not to ask for work, but suggest something you want to do, others will then tell you if that is possible and maybe help with the task. Packages can for now only be uploaded by Debian Developers, Ubuntu Developers or Debian Maintainers who contacted me directly and whose keys have been verified. This will be changed later, but at the current state of the Tanglu archive (= less safety checks for packages), I only want people to upload stuff who definitely have the knowledge to create sane packages (you can also proove that otherwise, of course). We will later establish a new-member process. If you want to provide a Tanglu archive mirror, we would be very happy, so that the main server doesn t have to carry all the load. If you have experience in creating Linux Live-CDs or have worked with the Ubiquity installer, helping with these parts would be awesome! Unfortunately, we cannot reuse parts of Linux Mint Debian, because many of their packages don t build from source and are repackaged binaries, which is a no-go for the Tanglu main archive. Sneak peek And here is a screenshot of the very first Tanglu installation (currently more Debian than Tanglu):

Something else I am involved in Debian for a very long time now, first as Debian Maintainer and then as Debian Developer and I never thought much about the work the Debian system administrators do. I didn t know how dak worked or how Wanna-build handles the buildds and what exactly the ftpmasters have to do. By not knowing, I mean I knew the very basic theory and what these people do. But this is something different than experiencing how much work setting up and maintaining the infrastructure is and what an awesome job the people do for Debian, keeping it all up and running and secure! Kudos for that, to all people maintaining Debian infrastructure! You rock! (And I will never ever complain about slow buildds or packages which stay in NEW for too long )

Update: This isn't actually that much better than letting them access the private key, since nothing is stopping the user from running their own SSH agent, which can be run under strace. A better solution is in the works. Thanks Timo Juhani Lindfors and Bob Proulx for both pointing this out. At work, we have a shared SSH key between the different people manning the support queue. So far, this has just been a file in a directory where everybody could read it and people would sudo to the support user and then run SSH. This has bugged me a fair bit, since there was nothing stopping a person from making a copy of the key onto their laptop, except policy. Thanks to a tip, I got around to implementing this and figured writing up how to do it would be useful. First, you need a directory readable by root only, I use /var/local/support-ssh here. The other bits you need are a small sudo snippet and a profile.d script. My sudo snippet looks like:

Defaults!/usr/bin/ssh-add env_keep += "SSH_AUTH_SOCK"
%support ALL=(root)  NOPASSWD: /usr/bin/ssh-add /var/local/support-ssh/id_rsa

Everybody in group support can run ssh-add as root. The profile.d goes in /etc/profile.d/support.sh and looks like:

if [ -n "$(groups   grep -E "(^  )support(  $)")" ]; then
    export SSH_AUTH_ENV="$HOME/.ssh/agent-env"
    if [ -f "$SSH_AUTH_ENV" ]; then
        . "$SSH_AUTH_ENV"
    fi
    ssh-add -l >/dev/null 2>&1
    if [ $? = 2 ]; then
        mkdir -p "$HOME/.ssh"
        rm -f "$SSH_AUTH_ENV"
        ssh-agent > "$SSH_AUTH_ENV"
        . "$SSH_AUTH_ENV"
    fi
    sudo ssh-add /var/local/support-ssh/id_rsa
fi

The key is unavailable for the user in question because ssh-add is sgid and so runs with group ssh and the process is only debuggable for root. The only thing missing is there's no way to have the agent prompt to use a key and I would like it to die or at least unload keys when the last session for a user is closed, but that doesn't seem trivial to do.

Following the DPL game call for players, here are my nominations for the fantastic four (in alphabetical order) :

1. Luca Falavigna
2. Tollef Fog Heen
3. Yves-Alexis Perez
4. Christian Perrier

These are four randomly selected people among those who share an understanding of the Debian community, a sense of leadership, and the ability to drive people forward with solutions.

Over the last couple of weeks, I have been working on getting binary packages for [Varnish] modules built. In the current version, you need to have a built, unpacked source tree to build a module against. This is being fixed in the next version, but until then, I needed to provide this in the build environment somehow. RPMs were surprisingly easy, since our RPM build setup is much simpler and doesn't use mock/mach or other chroot-based tools. Just make a source RPM available and unpack + compile that. Debian packages on the other hand, they were not easy to get going. My first problem was to just get the Varnish source package into the chroot. I ended up making a directory in /var/lib/sbuild/build which is exposed as /build once sbuild runs. The other hard part was getting Varnish itself built. sbuild exposes two hooks that could work: a pre-build hook and a chroot-setup hook. Neither worked: Pre-build is called before the chroot is set up, so we can't build Varnish. Chroot-setup is run before the build-dependencies are installed and it runs as the user invoking sbuild, so it can't install packages. Sparc32 and similar architectures use the linux32 tool to set the personality before building packages. I ended up abusing this, so I set HOME to a temporary directory where I create a .sbuildrc which sets $build_env_cmnd to a script which in turns unpacks the Varnish source, builds it and then chains to dpkg-buildpackage. Of course, the build-dependencies for modules don't include all the build-dependencies for Varnish itself, so I have to extract those from the Varnish source package too. No source available at this point, mostly because it's beyond ugly. I'll see if I can get it cleaned up.

Michael Biebl and I are giving a talk on systemd in Debian at FOSDEM on Sunday morning at 10. We'll be talking a bit about the current state in Wheezy, what our plans for Jessie are and what Debian packagers should be aware of. We would love to get input from people about what systemd in Jessie should look like, so if you have any ideas, opinions or insights, please come along. If you're just curious, you are also of course welcome to join.

gitano is not entirely unlike the non-web, server side of github. It allows you to create and manage users and their SSH keys, groups and repositories from the command line. Repositories have ACLs associated with them. Those can be complex ("allow user X to push to master in the doc/ subtree) or trivial ("admin can do anything"). Gitano is written by Daniel Silverstone, and I'd like to thank him both for writing it and for holding my hand as I went stumbling through my initial gitano setup. Getting started with Gitano can be a bit tricky, as it's not yet packaged and fairly undocumented. Until it is packaged, it's install from source time. You need luxio, lace, supple, clod, gall and gitano itself. luxio needs a make install LOCAL=1, the others will be installed to /usr/local with just make install. Once that is installed, create a user to hold the instance. I've named mine git, but you're free to name it whatever you would like. As that user, run gitano-setup and answer the prompts. I'll use git.example.com as the host name and john as the user I'm setting this up for. To create users, run ssh git@git.example.com user add john john@example.com John Doe, then add their SSH key with ssh git@git.example.com as john sshkey add workstation < /tmp/john_id_rsa.pub. To create a repository, run ssh git@git.example.com repo create myrepo. Out of the box, this only allows the owner (typically "admin", unless overridden) to do anything with it. To change ACLs, you'll want to grab the refs/gitano/admin branch. This lives outside of the space git usually use for branches, so you can't just check it out. The easiest way to check it out is to use git-admin-clone. Run it as git-admin-clone git@git.example.com:myrepo ~/myrepo-admin and then edit in ~/myrepo-admin. Use git to add, commit and push as normal from there. To change ACLs for a given repo, you'll want to edit the rules/main.lace file. A real-world example can be found in the NetSurf repository and the lace syntax might be useful. A lace file consists of four types of lines:

• Comments, start with -- or #
• defines, look like define name conditions
• allows, look like allow "reason" definition [definition ]
• denials, look like deny "reason" definition [definition ]

Rules are processed one by one, from the top and terminate whenever a matching allow or deny is found. Conditions can either be matches to an update, such as ref refs/heads/master to match updates to the master branch. To create groupings, you can use the anyof or allof verbs in a definition. Allows and denials are checked against all the definitions listed and if all of them match, the appropriate action is taken. Pay some attention to what conditions you group together, since a basic operation (is_basic_op, aka op_read and op_write) happens before git is even involved and you don't have a tree at that point, so rules like:

define is_master ref refs/heads/master
allow "Devs can push" op_is_basic is_master

simply won't work. You'll want to use a group and check on that for basic operations and then have a separate rule to restrict refs.

We recently switched from Buildbot to Jenkins at work, for building Varnish on various platforms. Buildbot worked-ish, but was a bit fiddly to get going on some platforms such as Mac OS and Solaris. Where buildbot has a daemon on each node that is responsible for contacting the central host, Jenkins uses SSH as the transport and centrally manages retries if a host goes down or is rebooted. All in all, we are pretty happy with Jenkins, except for one thing: The job configurations are a bunch of XML files and the way you are supposed to configure this is through a web interface. That doesn't scale particularly well when you want to build many very similar jobs. We want to build multiple branches, some which are not public and we want to build on many slaves. The latter we could partially solve with matrix builds, except that will fail the entire build if a single slave fails with an error that works on retry. As the number of slaves increases, such failures become more common. To solve this, I hacked together a crude tool that takes a yaml file and writes the XML files. It's not anywhere near as well structured and pretty as liw's jenkinstool, but it is quite good at translating the YAML into a bunch of XML files. I don't know if it's useful for anybody else, there is no documentation and so on, but if you want to take a look, it's on github. Feedback is most welcome, as usual. Patches even more so.

At work, we have a rotation of who is on call at a given time. We have few calls, but they do happen and so it's important to ensure both that a person is available, but also that they're aware they are on call (so they don't stray too far from their phone or a computer). In the grand tradition of abusing spreadsheets, we are using google docs for the roster. It's basically just two columns, one with date and one with user name. Since the volume is so low, people tend to be on call for about a week at a time, 24 hours a day. Up until now, we've just had a pretty old and dumb phone that people have carried around, but that's not really swish, so I have implemented a small system which grabs the current data, looks up the support person in LDAP and sends SMSes when people go on and off duty as well as reminding the person who's on duty once a day. If you're interested, you can look at the (slightly redacted) script.

Before I start, I'll admit that I'm not a real RPM packager. Maype I'm approaching this from completely the wrong direction, what do I know? I'm in the process of packaging Varnish 3.0.2 which includes mangling the spec file. The top of the spec file reads:

%define v_rc
%define vd_rc % ?v_rc:-% ?v_rc 

Apparently, this is not legal, since we're trying to define v_rc as a macro with no body. It's however not possible to directly define it as an empty string which can later be tested on, you have to do something like:

%define v_rc % nil 
%define vd_rc % ?v_rc:-% ?v_rc 

Now, this doesn't work correctly either. % ?macro tests if macro is defined, not whether it's an empty string so instead of two lines, we have to write:

%define v_rc % nil 
%if 0% ?v_rc  != 0
%define vd_rc % ?v_rc:-% ?v_rc 
%endif

The 0 ?v_rc != 0 workaround is there so that we don't accidentially end up with == 0 which would be a syntax error. I think having four lines like that is pretty ugly, so I looked for a workaround and figured that, ok, I'll just rewrite every use of % vd_rc to % ?v_rc:-% ?v_rc . There are only a couple, so the damage is limited. Also, I'd then just comment out the v_rc definition, since that makes it clear what you should uncomment to have a release candidate version. In my naivety, I tried:

# %define v_rc ""

# is used as a comment character in spec files, but apparently not for defines. The define was still processed and the build process stopped pretty quickly. Luckily, doing # % define "" seems to work fine and is not processed. I have no idea how people put up with this or if I'm doing something very wrong. Feel free to point me at a better way of doing this, of course.