c0d3 :: j0rg3

A collection of tips, tricks and snips. A proud Blosxom weblog. All code. No cruft.

Sat, 04 Mar 2017

Official(ish) deep dark onion code::j0rg3 mirror

Recently I decided that I wanted my blog to be available inside of the Deep, Dark Onion (Tor).

First time around, I set up a proxy that I modified to access only the clear web version of the blog and to avail that inside Tor as a ‘hidden service’.

My blog is hosted on equipment provided by the kind folk at insomnia247.nl and I found that, within a week or so, the address of my proxy was blocked. It’s safe for us to assume that it was simply because of the outrageous popularity it received inside Tor.

By “safe for us to assume” I mean that it is highly probable that no significant harm would come from making that assumption. It would not be a correct assumption, though.

What’s more true is that within Tor things are pretty durn anonymous. Your logs will show Tor traffic coming from 127.0.0.1 only. This is a great situation for parties that would like to scan sites repeatedly looking for vulnerabilities — because you can’t block them. They can scan your site over and over and over. And the more features you have (e.g., comments, searches, any form of user input), the more attack vectors are plausible.

So why not scan endlessly? They do. Every minute of every hour.

Since insomnia247 is a provider of free shells, it is incredibly reasonable that they don’t want to take the hit for that volume of traffic. They’re providing this service to untold numbers of other users, blogs and projects.

For that reason, I decided to set up a dedicated mirror.

Works like this: my blog lives here. I have a machine at home which uses rsync to make a local copy of this blog. Immediately thereafter it rsyncs any newly gotten data up to the mirror in onionland.

After consideration, I realized that this was also a better choice just in case there is something exploitable in my blog. Instead of even risking the possibility that an attacker could get access to insomnia247, they can only get to my completely disposable VPS which has hardly anything on it except this blog and a few scripts to which I’ve already opened the source code.

I’ve not finished combing through but I’ve taken efforts to ensure it doesn’t link back to clear web. To be clear, there’s nothing inherently wrong with that. Tor users will only appear as the IP address of their exit node and should still remain anonymous. To me, it’s just onion etiquette. You let the end-user decide when they want to step outside.

To that end, the Tor mirror does not have the buttons to share to Facebook, Twitter, LinkedIn, Google Plus.

That being said, if you’re a lurker of those Internet back-alleys then you can find the mirror at: http://aacnshdurq6ihmcs.onion

Happy hacking, friends!


Tags: , , , , , , , , , , ,
Permalink: 20170304.deep.dark.onion

Sat, 18 Feb 2017

The making of a Docker: Part II - Wickr: with bonus analysis

Recently, I read a rather excited attention-catching piece about how Wickr is the super-secure version of Slack. Attention caught in part because I feel like Wickr has been around for a while. I’d not seen anyone raving about its security in places where I normally interact with those who are highly informed about such subjects.

Good is that it seems the folk at Wickr did a fine job of making sure valuable data aren’t left behind.
The bad: closed-source, not subject to independent review; crazy marketin’-fancy-talk without a thorough description of how it does what is claimed.
Any time I’m looking at a product or service that boasts security, I sort of expect to see a threat model.

[ Update: At the time I was working on this project, the folk at Wickr were, evidently, opening their source. That’s spectacular news! Check it out on Github. ]

This began as an exercise to provide another piece of security-ish software in a Docker container. Anyone who has used a live distro (e.g., Kali, TAILS) with any regularity knows the ritual of installing favorite tools at each boot, data stores on removable media.

For me, there is tremendous appeal in reducing that to something like:
git clone https://georgeglarson/wickr
cd docker-wickr
./install.sh
wickr

Let’s dig in!

Having created a number of Docker containers my workflow is to queue up the base OS and go through the steps needed to get the software running while keeping careful notes. In this case, I had originally tried to install Wickr on a current copy of Kali. It was already known that Wickr, based off of Ubuntu 14.04, needed an older unicode library. So we begin with Ubuntu 14.04.

Grab a copy of Wickr and see what’s required:
dpkg -I wickr-me_2.6.0_amd64.deb

new debian package, version 2.0.
size 78890218 bytes: control archive=4813 bytes.
558 bytes, 14 lines control
558 bytes, 14 lines control64
10808 bytes, 140 lines md5sums
Package: wickr-me
Architecture: amd64
Section: net
Priority: optional
Version: 2.6.0-4
Replaces: wickr
Conflicts: wickr
Depends: libsqlcipher0, libuuid1, libicu52, libavutil52|libavutil54, libc6, libssl1.0.0, libx264-142, libglib2.0-0, libpulse0, libxrender1, libgl1-mesa-glx
Recommends: libnotify-bin, gstreamer-plugins0.10-good, gstreamer-plugins0.10-bad, gstreamer-plugins0.10-ugly
Maintainer: Wickr Inc.
Installed-Size: 200000
Description: Secure Internet Chat and Media Exchange agent
Wickr is a secure communications client

Okay. The CLI should do most of the work for us, giving a formatted list of dependencies.
dpkg -I wickr-me_2.6.0_amd64.deb | grep -E "^ Depends: | Recommends: " | sed -e "s/ Depends: //" -e "s/ Recommends: //" -e "s/,//g" -e "s/ / \\\ \n/g"

libsqlcipher0 \
libuuid1 \
libicu52 \
libavutil54 \
libc6 \
libssl1.0.0 \
libx264-142 \
libglib2.0-0 \
libpulse0 \
libxrender1 \
libgl1-mesa-glx
libnotify-bin \
gstreamer-plugins0.10-good \
gstreamer-plugins0.10-bad \
gstreamer-plugins0.10-ugly \

Attempting to get those with apt-get reports that it cannot find the gstreamer bits.

Let’s find:
apt-cache search gstreamer | grep -i plugin | grep -E "good|bad|ugly"

gstreamer0.10-plugins-good - GStreamer plugins from the "good" set
...
gstreamer0.10-plugins-bad - GStreamer plugins from the "bad" set
...
gstreamer0.10-plugins-ugly - GStreamer plugins from the "ugly" set

So, there’s the format we need to get the gstreamer dependencies. We know that we’ll also want SSH and wget. That should be enough for our Dockerfile.

We’ll pull down Wickr:
wget https://dls.wickr.com/Downloads/wickr-me_2.6.0_amd64.deb

Then install:
dpkg -i wickr-me_2.6.0_amd64.deb

Okay! We are, in theory, ready to run Wickr. We’re about to see we aren’t yet there — but these sorts of problems are pretty commonplace.
wickr-me

wickr-me: error while loading shared libraries: libxslt.so.1: cannot open shared object file: No such file or directory

Huh! We need libxslt. Let’s fix that: apt-get install libxslt1-dev

Now we can run it.
wickr-me

This application failed to start because it could not find or load the Qt platform plugin "xcb".

Available platform plugins are: eglfs, linuxfb, minimal, minimalegl, offscreen, xcb.

Reinstalling the application may fix this problem.
Aborted (core dumped)

One more: apt-get install xcb

Okay. That really was the last one. Now we have a complete list of dependencies for our Dockerfile:
RUN apt-get update && apt-get install -y \
gstreamer0.10-plugins-good \
gstreamer0.10-plugins-bad \
gstreamer0.10-plugins-ugly \
libsqlcipher0 \
libuuid1 \
libicu52 \
libavutil52 \
libc6 \
libssl1.0.0 \
libx264-142 \
libglib2.0-0 \
libpulse0 \
libxrender1 \
libxslt1-dev \
libgl1-mesa-glx \
libnotify-bin \
ssh \
wget \
xcb \
&& apt-get clean \

We now have Wickr in a Docker container and, because we are the curious sort, need to peek into what’s happening.

A natural first step is to set Wireshark atop Wickr. At a glance, seems to be communicating with a single IP address (204.232.166.114) via HTTPS.

Unsurprsingly, the client communicates to the server whenever a message is sent. Further it appears to poll the same address periodically asking for new messages. We see that the address resolves to Rackspace in San Antonio, TX.

We can easily establish the link between this IP address, Rackspace and the application.

Well, that’s enough. Right?

Good!

Wait.

What?

We’re still a little curious.

Aren’t we?

I mean, what’s the big question here? What happens if there’s a man in the middle? Persons so eagerly connect to any free WIFI, it is clearly a plausible scenario. Well… One way to find out!

Here’s what we learned. Server-side, the application is written in PHP. The IP address is resolved by the URI ‘secex.info’.

When we send, it calls ‘postMessage.php’:

When we receive, ‘downloadMessage.php’:

And it calls ‘newMessageCheck.php’ to, y’know, check for new messages.

Other analyses have forensically examined artefacts left behind; there are published descriptions of the encryption methods used for the local database connection. We didn’t go into more aggressive efforts such as disassembly because we are too lazy for that jazz!

My opinion, we didn’t learn anything wildly unexpected. Overall, Wickr seems an okay solution for convenient encrypted messaging. That’s always the trade: convenience vs. security. Least we ended with a Docker container for the software!

Github | Docker


Tags: , , ,
Permalink: 20170218.making.a.docker.wickr

Sun, 13 Jul 2014

Simple Protection with iptables, ipset and Blacklists

Seems I’ve always just a few more things going on than I can comfortably handle. One of those is an innocent little server holding the beginnings of a new project.

If you expose a server to the Internet, very quickly your ports are getting scanned and tested. If you’ve an SSH server, there are going to be attempts to login as ‘root’ which is why it is ubiquitously advised that you disable root login. Also why many advise against allowing passwords at all.

We could talk for days about improvements; it’s usually not difficult to introduce some form of two-factor authentication (2FA) for sensitive points of entry such as SSH. You can install monitoring software like Logwatch which can summarize important points from your logs, such as: who has logged via SSH, how many times root was used, etc.

DenyHosts and Fail2ban are very great ways to secure things, according to your needs.

DenyHosts works primarily with SSH and asks very little from you in way of configuration, especially if you’re using a package manager to install a version that is configured for the distribution on which you’re working. If you’re installing from source you may need to find where are your SSH logs (e.g., /var/log/secure, /var/log/auth.log). It’s extremely easy to set up DenyHosts to synchronize so that you’re automatically blocking widely-known offenders whether or not they’re after your server.

In contrast, Fail2ban is going to take more work to get set up. However, it is extremely configurable and works with any log file you point it toward which means that it can watch anything (e.g., FTP, web traffic, mail traffic). You define your own jails which means you can ban problematic IP addresses according to preference. Ban bad HTTP attempts from HTTP only or stick their noses in the virtual corner and don’t accept any traffic from them until they’ve served their time-out by completely disallowing their traffic. You can even use Fail2ban to scan its own logs, so repeating offenders can be locked out for longer.

Today we’re going to assume that you’ve a new server that shouldn’t be seeing any traffic except from you and any others involved in the project. In that case, you probably want to block traffic pretty aggressively. If you’ve physical access to the server (or the ability to work with staff at the datacenter) then it’s better to err in the direction of accidentally blocking good guys than trying to be overly fault-tolerant.

The server we’re working on today is a Debian Wheezy system. It has become a common misconception that Ubuntu and Debian are, intents and purposes, interchangeable. They’re similar in many respects and Ubuntu is great preparation for using Debian but they are not the same. The differences, I think, won’t matter for this exercise but I am unsure because this was written using Wheezy.

Several minutes after bringing my new server online, I started seeing noise in the logs. I was still getting set up and really didn’t want to stop and take protective measures but there’s no point in securing a server after its been compromised. The default Fail2ban configuration was too forgiving for my use. It was scanning for 10 minutes and banning for 10 minutes. Since only a few people should be accessing this server, there’s no reason for anyone to be trying a different password every 15 minutes (for hours).

I found a ‘close-enough’ script and modified it. Here, we’ll deal with a simplified version.

First, lets create a name for these ne’er-do-wells in iptables:
iptables -N bad_traffic

For this one, we’ll use Perl. We’ll look at our Apache log files to find people sniffing ‘round and we’ll block their traffic. Specifically, we’re going to check Apache’s ‘error.log’ for the phrases ‘File does not exist’ and ‘client denied by server configuration’ and block people causing those errors. This would be excessive for servers intended to serve the general populace. For a personal project, it works just fine as a ‘DO NOT DISTURB’ sign.


#!/usr/bin/env perl
use strict;
use POSIX qw(strftime);

my $log = ($ARGV[0] ? $ARGV[0] : "/var/log/apache2/error.log");
my $chain = ($ARGV[1] ? $ARGV[1] : "bad_traffic");

my @bad = `grep -iE 'File does not exist|client denied by server configuration' $log |cut -f8 -d" " | sed 's/]//' | sort -u`;
my @ablk = `/sbin/iptables -S $chain|grep DROP|awk '{print $4}'|cut -d"/" -f1`;

foreach my $ip (@bad) {
if (!grep $_ eq $ip, @ablk) {
chomp $ip;
`/sbin/iptables -A $chain -s $ip -j DROP`;
print strftime("%b %d %T",localtime(time))." badht: blocked bad HTTP traffic from: $ip\n";
}
}

That gives us some great, utterly unforgiving, blockage. Looking at the IP addresses attempting to pry, I noticed that most of them were on at least one of the popular block-lists.

So let’s make use of some of those block-lists! I found a program intended to apply those lists locally but, of course, it didn’t work for me. Here’s a similar program; this one will use ipset for managing the block-list though only minor changes would be needed to use iptables as above:

#!/bin/bash
IP_TMP=ip.tmp
IP_BLACKLIST_TMP=ip-blacklist.tmp

IP_BLACKLIST=ip-blacklist.conf

WIZ_LISTS="chinese nigerian russian lacnic exploited-servers"

BLACKLISTS=(
"http://danger.rulez.sk/projects/bruteforceblocker/blist.php" # BruteForceBlocker IP List
"http://rules.emergingthreats.net/blockrules/compromised-ips.txt" # Emerging Threats - Compromised IPs
"http://www.spamhaus.org/drop/drop.txt" # Spamhaus Don't Route Or Peer List (DROP)
"http://www.spamhaus.org/drop/edrop.txt" # Spamhaus Don't Route Or Peer List (DROP) Extended
"http://cinsscore.com/list/ci-badguys.txt" # C.I. Army Malicious IP List
"http://www.openbl.org/lists/base.txt" # OpenBL.org 90 day List
"http://www.autoshun.org/files/shunlist.csv" # Autoshun Shun List
"http://lists.blocklist.de/lists/all.txt" # blocklist.de attackers
)

for address in "${BLACKLISTS[@]}"
do
echo -e "\nFetching $address\n"
curl "$address" >> $IP_TMP
done

for list in $WIZ_LISTS
do
wget "http://www.wizcrafts.net/$list-iptables-blocklist.html" -O - >> $IP_TMP
done

wget 'http://wget-mirrors.uceprotect.net/rbldnsd-all/dnsbl-3.uceprotect.net.gz' -O - | gunzip | tee -a $IP_TMP

grep -o '^[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}[/][0-9]\{1,3\}' $IP_TMP | tee -a $IP_BLACKLIST_TMP
grep -o '^[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}[^/]' $IP_TMP | tee -a $IP_BLACKLIST_TMP

sed -i 's/\t//g' $IP_BLACKLIST_TMP
sort -u $IP_BLACKLIST_TMP | tee $IP_BLACKLIST

rm $IP_TMP
rm $IP_BLACKLIST_TMP

wc -l $IP_BLACKLIST

if hash ipset 2>/dev/null
then
ipset flush bloxlist
while IFS= read -r ip
do
ipset add bloxlist $ip
done < $IP_BLACKLIST
else
echo -e '\nipset not found\n'
echo -e "\nYour bloxlist file is: $IP_BLACKLIST\n"
fi


Download here:
    bad_traffic.pl
    bloxlist.sh


Tags: , , , , , , , , , ,
Permalink: 20140713.simple.protection.with.iptables.ipset.and.blacklilsts

Thu, 13 Jun 2013

Blogitechture continued… Simplify with Vim

Last we were discussing the structure and design of your own CLI-centric blog platform, we had some crude methods of starting and resuming posts before publishing.

Today, let’s explore a little more into setting up a bloging-friendly environment because we need to either make the experience of blogging easy or we’ll grow tired of the hassle and lose interest.

We can reasonably anticipate that we won’t want to beleaguered with repetitious typing of HTML bits. If we’re going to apply paragraph tags, hyperlinks, codeblocks, etc. with any frequency, that task is best to be simplified. Using Vim as our preferred editor, we will use Tim Pope’s brilliant plug-ins ‘surround’ and ‘repeat’, combined with abbreviations to take away the tedium.

The plug-ins just need dropped into your Vim plugin directory (~/.vim/plugin/). The directory may not exist if you don’t have any plug-ins yet. That’s no problem, though. Let’s grab the plugins:

cd ~/.vim/
wget "http://www.vim.org/scripts/download_script.php?src_id=19287" -O surround.zip
wget "http://www.vim.org/scripts/download_script.php?src_id=19285" -O repeat.zip

Expand the archives into the appropriate directories:

unzip surround.zip
unzip repeat.zip

Ta-da! Your Vim is now configured to quickly wrap (surround) in any variety of markup. When working on a blog, you might use <p> tags a lot by putting your cursor amid the paragraph and typing yss<p>. The plug-in will wrap it with opening and closing paragraph tags. Move to your next paragraph and then press . to repeat.

That out of the way, let’s take advantage of Vim’s abbreviations for some customization. In our .vimrc file, we can define a few characters that Vim will expand according to their definition. For example, you might use:
ab <gclb> <code class="prettyprint lang-bsh linenums:1">
Then, any time you type <gclb> and bress <enter>, you’ll get:
<code class="prettyprint lang-bsh linenums:1">

The next time that we take a look at blogitecture, we will focus on making the posts convenient to manage from our CLI.


Tags: , , , ,
Permalink: 20130613.blogitechture.continued

Sun, 09 Jun 2013

ixquick link maker

In an effort to promote practical privacy measures, when I send people links to search engines, I choose ixquick. However, my personal settings submit my search terms via POST data rather than GET, meaning that the search terms aren’t in the URL.

Recently, I’ve found myself hand-crafting links for people and then I paste the link into a new tab, to make sure I didn’t fat-finger anything. Not a problem per se, but the technique leaves room for a bit more efficiency. So I’ve taken the ‘A Search Box on Your Website’ tool offered by ixquick and slightly modified the code it offers, to use GET variables, in a new tab where I can then copy the URL and provide the link to others.

You can test, or use, it here — I may add it (or a variant that just provides you the link) to the navigation bar above. First, though, I’m going to mention the need to the outstanding minds at ixquick because it would make a LOT more sense on their page than on mine.


Tags: ,
Permalink: 20130609.ixquick.search