<?xml version="1.0" encoding="utf-8" standalone="yes"?><rssversion="2.0"xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>1533B4dC0.de</title><link>https://www.1533b4dc0.de/</link><description>1533B4dC0.de</description><generator>Hugo -- gohugo.io</generator><language>en-us</language><lastBuildDate>Thu, 24 Feb 2022 00:00:00 +0000</lastBuildDate><atom:linkhref="https://www.1533b4dc0.de/index.xml"rel="self"type="application/rss+xml"/><item><title>About me</title><link>https://www.1533b4dc0.de/about/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://www.1533b4dc0.de/about/</guid><description><p>My name&rsquo;s Peter. I&rsquo;m a passionate software developer especially interested in all kind of networking stuff but also asynchronous data processing, software architecture, testing and automatic software quality analysis and many more.</p>
<p>I&rsquo;m the author of <a href="https://gitlab.com/inetmock/inetmock">InetMock</a> and <a href="https://github.com/baez90/goveal">Goveal</a> (more on <a href="https://www.1533b4dc0.de/projects">projects</a>) but I&rsquo;m also trying to contribute to other open source projects.</p></description></item><item><title>Projects</title><link>https://www.1533b4dc0.de/projects/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://www.1533b4dc0.de/projects/</guid><description><h2 id="inetmock" >INetMock
</h2><p><a href="https://gitlab.com/inetmock/inetmock">INetMock</a> started as an resource/container friendly alternative to <a href="https://www.inetsim.org/">INetSim</a>.
While working on a project we tried to reduce analysis complexity coming from &rsquo;noise&rsquo; in the network traffic recorded to a central INetSim cluster we were running.
We decided to decentralize the internet simulation, put it into a container image and run directly on every host multiple times in virtual networks.
Unfortunately INetSim has a relatively huge memory footprint (~1GB) which alone wouldn&rsquo;t been a showstopper but in combination with a relatively long startup time I felt having something smaller could be beneficial so I started to implement a prototype in Go.</p>
<p>2 years later INetMock has grown to kind of a full router (supporting DNS and DHCP) with support for faking HTTP/s (direct or proxy requests) requests.
Furthermore it is able to record PCAP files for further analysis and it emits events for every handled request.</p>
<p>It comes with a descriptive configuration language (embedded in a YAML configuration) to setup the behavior of all components and to define health checks/integration tests to validate your configuration.</p>
<p>Apart from working as a router it can also be used e.g. for integration tests of HTTP APIs, DNS/DoT/DoH clients and most likely other things I haven&rsquo;t even thought about.</p>
</h2><p><a href="https://github.com/baez90/goveal">Goveal</a> is similar to <a href="https://github.com/webpro/reveal-md">reveal-md</a> or previously <em>GitPitch</em> but obviously in Go.
Originally I used GitPitch but then the author decided to go with a commercial license.
The commercial license made sense when I was working at the university but after that it didn&rsquo;t really make sense any more.
So I decided to replace it with a small custom CLI rendering the markdown into a static HTML file and serving it as a local web server (basically).</p>
Currently I&rsquo;m working on a rewrite which adds e.g. 1st class support for <a href="https://mermaid-js.github.io">mermaid-js</a> diagrams in slides.</p></description></item><item><title>Libvirt & Podman: network 'mesh'</title><link>https://www.1533b4dc0.de/post/libvirt-podman-network-mesh/</link><pubDate>Thu, 24 Feb 2022 00:00:00 +0000</pubDate><guid>https://www.1533b4dc0.de/post/libvirt-podman-network-mesh/</guid><description><p><em>Disclaimer: I tested all this with Podman 3.x even though Podman 4.0 is already announced but the CNI driver is still available with Podman 4.0 and as soon as I get my hands on 4.0 I&rsquo;ll give <strong>Netavark</strong> a try, too!</em></p>
<p>When playing around with containers and VMs one might ask if it&rsquo;s possible to bring VMs and containers into a common network segment.
I see &lsquo;why the hell would I need a VM anyway when already having containers&rsquo; or something similar I almost see on your face 😜</p>
<p>Well 1st of all, not everything can be solved with containers.
For instance windows applications can be run in Windows containers but I&rsquo;m not aware of how to run a Windows container on my Linux desktop.</p>
<p>But also in pure Linux environments there are cases where a VM is probably a better fit for the problem.
As you might know I&rsquo;m a bit of network 🤓 and I love playing around with &lsquo;weird&rsquo; stuff almost no one else does even think about if not forced to.
So if you try to implement for example your own DHCP server you might want to isolate your experiments (especially at the beginning) to avoid discussion about &ldquo;why&rsquo;s Netflix on the TV not working?!&rdquo; 😄 or also if you try to implement your own &lsquo;firewall&rsquo; with DNAT support (stay tuned - post&rsquo;s following!).</p>
</h2><p>Okay now that I came around with <em>some</em> arguments - if they&rsquo;re convincing or not is not important - how does this work?</p>
<p>Assuming you&rsquo;ve Libvirt and Podman already installed on your system without any modification and you run</p>
<li>the <code>&lt;mac/&gt;</code> for the bridge interface (of the host)</li>
<li>the <code>&lt;ip/&gt;</code> of the host on the bridge interface
<ul>
<li>an <em>optional</em> <code>&lt;dhcp/&gt;</code> range definition</li>
</ul>
</li>
</ul>
<p>The complete reference for the XML schema can be found <a href="https://libvirt.org/formatnetwork.html">here</a>.</p>
<p>Before we have a closer look how to bring Podman containers into a Libvirt network, let&rsquo;s define a new <code>containers</code> network.
The following snippet contains the definition I&rsquo;ll use:</p>
<li>remove the <code>&lt;forward/&gt;</code> block</li>
<li>change the <code>&lt;name/&gt;</code> and the <code>&lt;uuid/&gt;</code> (with the help of <code>uuidgen</code>)</li>
<li>change the <code>name=&quot;&quot;</code> of the <code>&lt;bridge/&gt;</code></li>
<li>change the <code>address=&quot;&quot;</code> attribute of the <code>&lt;mac/&gt;</code> (use any <a href="https://macaddress.io/mac-address-generator">mac address generator</a>)</li>
<li>change the <code>address=&quot;&quot;</code> attribute of the <code>&lt;ip/&gt;</code> and <code>start=&quot;&quot;</code> and <code>end=&quot;&quot;</code> of the DHCP range accordingly</li>
</ul>
<p>You may use any private network - as far as I can tell it shouldn&rsquo;t matter if you&rsquo;re using a class B, C or D private network as long as you don&rsquo;t have any conflicts with your LAN or any other virtual interfaces of your environment.</p>
<p>When done safe your network definition as <code>.xml</code> file.
To import the configuration you can use <code>virsh net-define</code> like in the following snippet (assuming the network definition is in <code>containers.xml</code>):</p>
</span></span><span style="display:flex;"><span>&gt; Network containers defined from containers.xml
</span></span></code></pre></div><p><em>Note: this only works because the XML already contains an <code>&lt;uuid/&gt;</code>. Otherwise you&rsquo;d have to use <code>virsh net-create</code> and a few more extra steps to make the network actually persistent.</em></p>
<p>If you now check with <code>virsh net-list</code> you&rsquo;d be disappointed because there&rsquo;s no network!
Checking again with <code>virsh net-list --all</code> explains why our <code>containers</code> network wasn&rsquo;t in the output previously because it is by default <em>inactive</em>.
To activate it we&rsquo;ve to start it like so:</p>
</span></span><span style="display:flex;"><span>&gt; Network containers started
</span></span></code></pre></div><p>If you don&rsquo;t mind the extra adapter and wish to use the network frequently in the future you might consider to autostart it:</p>
</h2><p><em>Note: this only works with <strong>rootfull</strong> Podman because rootless Podman does not use CNI but another network stack.</em></p>
<p>A clean Podman installation without any custom network created comes with the default network <code>podman</code>.
Rootfull Podman network configs are by default stored in <code>/etc/cni/net.d</code>.
You should find the default network as <code>87-podman.conflist</code> in the aforementioned directory.</p>
<p>Every Podman network is defined as JSON file.
We will define our own <code>libvirt</code> network to join Podman containers into the previously created Libvirt network.
You can either use <code>podman network create</code> to create the network (at least more or less) or you can copy for example the default network and make some adjustments.</p>
<p>To create the new network from the CLI you can use the following command:</p>
</span></span></code></pre></div><p>Note that I used a different IP range as in the Libvirt network! Otherwise Podman will complain that the IP range is already in use at an adapter.
You can use this command to create the required file in <code>/etc/cni/net.d/</code> but you&rsquo;ve to update the <code>ranges</code> accordingly before creating a container in the network.</p>
<p>Because we&rsquo;ve to edit the <code>.conflist</code> either way copy the default one is also fine.</p>
<p>The <code>.conflist</code> I&rsquo;m using looks like this:</p>
</span></span></code></pre></div><p>Interestingly the <code>rangeStart</code> and <code>rangeEnd</code> are actually IP addresses and not tight to some IP networks but unfortunately there&rsquo;s no equivalent for <code>podman network create</code> hence I update both to a range after the DHCP range of the Libvirt network to make sure that no duplicate IPs are assigned.</p>
<p>I tend to declare the network as <code>host-local</code> but this shouldn&rsquo;t be critical.
The <strong>most important</strong> part is to update the <code>bridge</code> to the same interface like in the Libvirt network definition (in my case <code>conbr0</code>).</p>
<p>After this we&rsquo;re ready to go and you can for instance start a Nginx container in the <code>libvirt</code> network and you should be able to reach it from a VM in the Libvirt network:</p>
</span></span></code></pre></div><p>A nice option for <code>podman run</code> is <code>--ip</code>.
You&rsquo;ve to choose an IP from the previously configured <code>range</code> but you can skip the <code>podman inspect</code> or <code>ip a</code> to get the container IP and the container will have the IP every time you start it, if you like 😉 and speaking of &rsquo;nice&rsquo; <code>podman run</code> options: you do know <code>--replace</code>, don&rsquo;t you?</p></description></item></channel></rss>
