<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Sam Decrock on Medium]]></title>
        <description><![CDATA[Stories by Sam Decrock on Medium]]></description>
        <link>https://medium.com/@samdecrock?source=rss-8f8336fa1565------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*4sPWlLk6QhGMke6-.jpeg</url>
            <title>Stories by Sam Decrock on Medium</title>
            <link>https://medium.com/@samdecrock?source=rss-8f8336fa1565------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 06:46:12 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@samdecrock/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Bulk convert HEIC to JPG, on a Mac, really easy]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://samdecrock.medium.com/bulk-convert-heic-to-jpg-on-a-mac-really-easy-35c2103df8e9?source=rss-8f8336fa1565------2"><img src="https://cdn-images-1.medium.com/max/600/1*Cs1tQtiC40Xf9PL-MRuptQ.png" width="600"></a></p><p class="medium-feed-snippet">Getting your iPhone images onto your Mac is easy, but once there, there&#x2019;s not much you can do with them because, since iOS 11, all iPhone&#x2026;</p><p class="medium-feed-link"><a href="https://samdecrock.medium.com/bulk-convert-heic-to-jpg-on-a-mac-really-easy-35c2103df8e9?source=rss-8f8336fa1565------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://samdecrock.medium.com/bulk-convert-heic-to-jpg-on-a-mac-really-easy-35c2103df8e9?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/35c2103df8e9</guid>
            <category><![CDATA[heic]]></category>
            <category><![CDATA[heic-to-jpg]]></category>
            <category><![CDATA[terminal]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Thu, 22 Dec 2022 08:36:32 GMT</pubDate>
            <atom:updated>2022-12-22T08:37:22.037Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Building Electron.js apps with React.js (2023 update)]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://samdecrock.medium.com/building-electron-js-apps-with-react-js-2022-4d14fb2924ac?source=rss-8f8336fa1565------2"><img src="https://cdn-images-1.medium.com/max/1824/1*qdlKAYsBa-i3wt3VxtOFOQ.png" width="1824"></a></p><p class="medium-feed-snippet">This tutorial explains how you can develop and build your Electron.js app using React.js</p><p class="medium-feed-link"><a href="https://samdecrock.medium.com/building-electron-js-apps-with-react-js-2022-4d14fb2924ac?source=rss-8f8336fa1565------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://samdecrock.medium.com/building-electron-js-apps-with-react-js-2022-4d14fb2924ac?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/4d14fb2924ac</guid>
            <category><![CDATA[reactjs]]></category>
            <category><![CDATA[create-electronjs-app]]></category>
            <category><![CDATA[electronjs]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Wed, 21 Dec 2022 01:15:23 GMT</pubDate>
            <atom:updated>2023-04-04T08:27:22.975Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Transferring data between Kubernetes clusters]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://samdecrock.medium.com/transferring-data-between-kubernetes-clusters-c0c1b59930d1?source=rss-8f8336fa1565------2"><img src="https://cdn-images-1.medium.com/max/1446/1*lnJLID-wi0tGMX2HIAGTXA.png" width="1446"></a></p><p class="medium-feed-snippet">With this simply trick you can easily move data between Kubernetes clusters.</p><p class="medium-feed-link"><a href="https://samdecrock.medium.com/transferring-data-between-kubernetes-clusters-c0c1b59930d1?source=rss-8f8336fa1565------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://samdecrock.medium.com/transferring-data-between-kubernetes-clusters-c0c1b59930d1?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/c0c1b59930d1</guid>
            <category><![CDATA[persistent-volume]]></category>
            <category><![CDATA[kubernetes-cluster]]></category>
            <category><![CDATA[persistent-volume-claim]]></category>
            <category><![CDATA[kubernetes]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Wed, 14 Dec 2022 21:02:58 GMT</pubDate>
            <atom:updated>2022-12-27T20:55:04.496Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Getting a smart electric boiler — The Ariston Velis]]></title>
            <link>https://samdecrock.medium.com/getting-a-smart-electric-boiler-d8029ec7cc87?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/d8029ec7cc87</guid>
            <category><![CDATA[electric-boiler]]></category>
            <category><![CDATA[ariston]]></category>
            <category><![CDATA[ariston-evo-dry-wifi]]></category>
            <category><![CDATA[electric-boilers-market]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Mon, 17 Oct 2022 18:56:15 GMT</pubDate>
            <atom:updated>2022-12-19T14:35:28.103Z</atom:updated>
            <content:encoded><![CDATA[<p>It started as a simple task: find a smart boiler with wifi. Why? Because I want to make it smarter by attaching it to my PV panels and have it heat up more water when the solar panels produce extra electricity.</p><p>I quickly stumbled upon the Ariston Velis Evo Dry. It’s a smart boiler which learns your hot water use behaviour and uses that to only heat the water when needed. Not only does it look good, it’s only 27 cm thick and it looks beautiful. It has 2 tanks which only fill up if you really need them (being smart here) and incoming cold water is slowly mixed with hot water of the tank using some kind of layered technology called Nanomix. The manual even said something about Wifi so that looked like the perfect fit for me. I can’t wait to have a look at its APIs 😁.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E4dN-4Gzbx6K7bYu-dERaw.png" /><figcaption>Doesn’t it look beautiful? I know, it’s just a boiler. But is has Wifi!</figcaption></figure><p>However, I quickly found out that the Wifi wasn’t that self-evident, nor was the ‘Dry’ part of the boiler. I also like the metal brushing (or is it grey plastic?), but even that was not self-evident.</p><p>I found different model numbers for what seemed to be the same machine. Selecting different countries on the Ariston website also results in different models so that looked like hell to me.</p><p>For myself, and for others that want the same smart boiler, here’s the full list of models I found searching the web:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/99b8ca5b63d3dfdf0d8a4df1ba84d43d/href">https://medium.com/media/99b8ca5b63d3dfdf0d8a4df1ba84d43d/href</a></iframe><p>You may think I’m crazy, compiling this list. But this gives a big insight into which models have been made over the years. Note that the dates you see are extracted from the pdf’s metadata.</p><p>When looking at the list, you will see that they essentially have 7 models:</p><ul><li>Velis Evo Inox (2015)</li><li>Velis Evo (2016)</li><li>Velis Evo Wifi (2016)</li><li>Velis Evo Plus (2016)</li><li>Velis Evo Plus Wifi (2017)</li><li>Velis Evo Dry (2019)</li><li>Velis Evo Dry Wifi (2019)</li></ul><p>The weird thing though is that some models seem to have 2 model numbers. For example the Velis Evo Dry with a capacity of 80 liters has model numbers 3626266 and 3626245 . The first is marketed as the Velis Evo Dry 100 EU and the second one as the Velis Evo Dry 80 FR EU. Checking the datasheet, the second one has 2 resistors per tank while the first one only has one. This makes the boiler slightly heavier. The reason the earlier version was marketed as 100 while its real content is 80 liters is because it compares to a 100 liters traditional boiler.</p><p>Also, the UK model comes with an extra unvented kit while the European models don’t. That must be some kind of regulatory thing.</p><p>Obviously I’m interested in the ‘dry’ version. ‘Dry’ means the resistor doesn’t come in direct contact with the water which prevents limescale. I must say there are some online shops out there who wrongly market the Velis Evo Plus as the Velis Evo Dry which only adds to the confusion 😔.</p><p>Living in Belgium, I can buy the latest Velis Evo Dry at a <a href="https://blue.vanmarcke.com/nl_BE/Verwarming-en-warm-water/Warmwaterbereiders/Elektrische-boilers/droge-weerstand/Ariston---Velis-Evo-Dry---elektrische-boiler/p/293050">local webshop</a>. Unfortunalty <a href="https://www.ariston.com/cms/s3Downloader/?sku=velis_evo_dry&amp;locale=fr-fr&amp;file=Velis%20Evo%20Dry-%20Ariston-Documentation.pdf">the wifi version</a> is only available in France. Curiously the iOS app is also not available on the Belgian app store.</p><p>I guess I’ll either have to settle for the non-wifi version or cross the border and get one in France.</p><h3><strong>Update: The Velis 3.0</strong></h3><p>As I really wanted the wifi version, I checked with several French webshops and they all told me the same: the wifi version isn’t available anymore. I contacted Ariston France and they told me they don’t produce it anymore for “commercial purposes”.</p><p>I also contacted Ariston Belgium and they responded that because of “strategic choices” the Velis Wifi will only be available in 2024 in Belgium!</p><p>I then stumbled upon <a href="https://www.aristongroup.com/media/files/4378_Ariston%20Group_FY2021_Annual%20Report.pdf">Ariston’s anual report of 2021</a>. In that report I discovered that they created a new and improved Velis, called the Velis 3.0. But, and this most surprising, they presented it in the summer of 2021 in Russia! Indeed, looking up the Velis 3.0, it’s for sale on <a href="https://ariston-pro.com/ru/catalog/vodonagrevateli-elektricheskie/nakopitelnye/velis-lux-inox-power-abse-wifi/">Russians websites</a> as well as on <a href="https://100kotlov.kiev.ua/p/1267974094-boyler-ariston-velis-lux-pw-abse-dry-wifi-100/">Ukrainian sites</a>. Of all countries, these are the only 2 countries in the world that sell the latest and greatest smart boiler 😮</p><p>I must admit, it looks damn good:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/725/1*BRCH1yIGY0CUUlsAg4JBwQ.jpeg" /></figure><p>Plus, they also have it in the 100 litres version, which is the same height als the older 80 litres version.</p><p>That might explain why the Velis Wifi is due for 2024 in Belgium. They probably have a serious problem with the current Russian sanctions as the Velis 3.0 is being produced in St. Petersburg, Russia.</p><h3>Is it worth the money?</h3><p>I was thinking of buying a cheap boiler until the wifi version comes out in 2024, but doing the math reveals I should buy the Velis anyway. The Velis Evo Dry 80 liters can be bought for around € 600 while a cheap 100 liter boiler is about half the price. Yet, the annual power consumption doubles with the cheap boiler (2703 kWh/year vs 1291 kWh/year). That’s around 1400 kWh more per year. At a price of € 0,50/kWh that’s a saving of € 700 per year. With those savings I can buy a new smart boiler next year 🙂. Even when electrical costs drop back to € 0,10/kWh (as in 2021) we still save money after 2–3 years.</p><h3>Can I take a bath?</h3><p>Throughout this investigation, I always wondered if the 80 liters is enough to take a bath. As they’re always taking about showers, it’s pretty confusing if they even advise to use the Velis for bath taking.</p><p>Doing the math it should be possible. Mixing 80 liters of hot water at 80°C with 90 liters of cold water at 5°C would result in 170 liters of 40°C which is more than enough to fill a bath and even take some extra showers. Off course, you don’t want the water to be that hot all the time.</p><p>I even found this guide which states that the 80 liters Velis would provide enough water for taking a bath, a shower, use the bathroom sink and even do the dishes:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/878/1*IgD2_B1UvV_ygf5JnsVY2Q.png" /><figcaption>Guide on the side of the box of the Velis Evo Dry 100 EU (model 3626266)</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d8029ec7cc87" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Patching Microsoft’s RDP service yourself]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://samdecrock.medium.com/patching-microsofts-remote-desktop-service-yourself-db25a4d8bc64?source=rss-8f8336fa1565------2"><img src="https://cdn-images-1.medium.com/max/888/1*Hnr5nMYMydrlDaW0jr0ONw.png" width="888"></a></p><p class="medium-feed-snippet">I&#x2019;ll explain how patching termsrv.dll will give you more than one RDP connection.</p><p class="medium-feed-link"><a href="https://samdecrock.medium.com/patching-microsofts-remote-desktop-service-yourself-db25a4d8bc64?source=rss-8f8336fa1565------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://samdecrock.medium.com/patching-microsofts-remote-desktop-service-yourself-db25a4d8bc64?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/db25a4d8bc64</guid>
            <category><![CDATA[hacking]]></category>
            <category><![CDATA[rdp-services]]></category>
            <category><![CDATA[disassembly]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Fri, 05 Aug 2022 10:47:03 GMT</pubDate>
            <atom:updated>2023-03-22T07:41:25.171Z</atom:updated>
        </item>
        <item>
            <title><![CDATA[Use Kubernetes’ ExtendedResourceToleration to keep out non-GPU pods]]></title>
            <link>https://medium.com/robovision-ai/use-kubernetes-extendedresourcetoleration-to-keep-out-non-gpu-pods-c180f5c5e76d?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/c180f5c5e76d</guid>
            <category><![CDATA[kubernetes-cluster]]></category>
            <category><![CDATA[kubernetes-api]]></category>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[taints-and-toleration]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Wed, 20 Oct 2021 14:42:19 GMT</pubDate>
            <atom:updated>2024-05-13T11:42:55.969Z</atom:updated>
            <content:encoded><![CDATA[<p>For all those using GKE, you might have noticed Google does a pretty good job at managing GPU pods. That is, pods that request a GPU automatically get assigned a GPU node and non-GPU-pods are kept out.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*p_IpKYuHT0PceESpbGDmaA.jpeg" /></figure><p>From <a href="https://cloud.google.com/kubernetes-engine/docs/how-to/gpus#create">the GKE docs</a>:</p><blockquote><em>When you add a GPU node pool to an existing cluster that already runs a non-GPU node pool, GKE automatically </em><a href="https://cloud.google.com/kubernetes-engine/docs/how-to/node-taints"><em>taints</em></a><em> the GPU nodes with the following node taint:</em></blockquote><blockquote><em>nvidia.com/gpu:NoSchedule</em></blockquote><blockquote><em>Additionally, GKE automatically applies the corresponding tolerations to Pods requesting GPUs by running the </em><a href="https://kubernetes.io/docs/reference/access-authn-authz/admission-controllers/#extendedresourcetoleration"><em>ExtendedResourceToleration</em></a><em> admission controller.</em></blockquote><blockquote><em>This causes only Pods requesting GPUs to be scheduled on GPU nodes, which enables more efficient autoscaling: your GPU nodes can quickly scale down if there are not enough Pods requesting GPUs.</em></blockquote><p>That may sound like something pretty cool. So cool we also wanted this on our own clusters here at Robovision. Because we use some heavy <a href="https://www.ray.io/">Ray Workers</a> for our trainings, we need to keep those GPU nodes free for GPU pods only. We don’t want someone’s pet project taking up resources when those resources can be used to do our Robovision magic: training new models.</p><h3>Taints, tolerations, extended resources and admission controllers</h3><p>There are a bunch of things you need to understand before we can move the ultimate solution. But don’t worry, I’ll explain them here.</p><h4>Taints &amp; Tolerations</h4><p>If you add a <a href="https://kubernetes.io/docs/concepts/scheduling-eviction/taint-and-toleration/">taint</a> to a node, only pods with matching <a href="https://kubernetes.io/docs/concepts/scheduling-eviction/taint-and-toleration/">tolerations</a> are allowed to run on those nodes. Let’s say you add the following taint to a node indicating only Trekkies are allowed on that node (couldn’t think of a better joke, sorry):</p><pre>kubectl taint nodes somenode trekkies-only=:NoSchedule</pre><p>Only pods with the following toleration are allowed to run on those nodes:</p><pre>tolerations:<br>- key: &quot;trekkies-only&quot;<br>  operator: &quot;Exists&quot;<br>  effect: &quot;NoSchedule&quot;</pre><p>This prevents those pesky Star Wars pods from running on our node 😜</p><p>A more practical example is adding the following taint indicating only pods really needing a GPU can run on that node:</p><pre>kubectl taint nodes somenode nvidia.com/gpu=:NoSchedule</pre><p>If you want your pod to run on that node, you will need to add the following toleration to that pod:</p><pre>tolerations:<br>  - key: &quot;nvidia.com/gpu&quot;<br>    operator: &quot;Exists&quot;<br>    effect: &quot;NoSchedule&quot;</pre><p>Yes, this already prevents other pods from running on that node, but adding this piece of yaml to your manifest can be cumbersome, especially if you don’t have control over the pod spec or if you don’t want to annoy your data scientists with adding those extra tolerations.</p><h4>Extended resources</h4><p>Since you are using GPUs in your cluster, this means you’ll probably be using the extended resource called nvidia.com/gpu.</p><p>nvidia.com/gpu is called and <a href="https://kubernetes.io/docs/tasks/configure-pod-container/extended-resource/">extended resource</a> because it is not a default kubernetes resources like cpu and memory. In fact you can define any extended resource as long as it adheres to the following format: example.com/foo</p><p>For example, the following would be a valid resource request:</p><pre>resources:<br>  requests:<br>    cpu: 1<br>    memory: 100Mi<br>    robovision.ai/monkeys: 3</pre><p>robovision.ai/monkeys being the extended resource here. And we want 3 of them because monkeys bring joy to our pods!</p><h4>Admission controller: ExtendedResourceToleration</h4><p>This brings us neatly to the admission controller called ExtendedResourceToleration. What’s an <a href="https://kubernetes.io/docs/reference/access-authn-authz/admission-controllers/">admission controller</a> you may ask? It’s a Kubernetes plugin that intercepts requests going to the Kubernetes API. If you kubectl apply something, such an admission controller may kick in and act upon that request. One of those admission controllers is the ExtendedResourceToleration. The ExtendedResourceToleration is a very nifty plugin that checks for incoming extended resources and magically turns those into a Toleration. This means if the extended resource robovision.ai/monkeys: 3 comes in, it adds the following toleration to that pod:</p><pre>tolerations:<br>  - key: &quot;robovision.ai/monkeys&quot;<br>    operator: &quot;Exists&quot;<br>    effect: &quot;NoSchedule&quot;</pre><p>Or if the more useful extended resource nvidia.com/gpu comes in, the ExtendedResourceToleration plugin would add the following toleration to the pod:</p><pre>tolerations:<br>  - key: &quot;nvidia.com/gpu&quot;<br>    operator: &quot;Exists&quot;<br>    effect: &quot;NoSchedule&quot;</pre><p>This means that if you have a node tainted with nvidia.com/gpu=:NoSchedule your GPU requesting pods will automatically get the right toleration and will therefor be allowed to run on the node. Because non-gpu pods don’t have this nvidia.com/gpu resource, they won’t get this toleration and therefor won’t be allowed on that node.</p><p>This is exactly what GKE (and <a href="https://github.com/Azure/AKS/blob/master/CHANGELOG.md#release-2020-10-19">recently AKS</a>) is doing when you add a GPU pod.</p><h3>How to enable the ExtendedResourceToleration</h3><p>If you’re using <a href="https://kubernetes.io/docs/reference/setup-tools/kubeadm/kubeadm-init/">kubeadm</a> like we do, it’s fairly simple the specify the ExtendedResourceToleration plugin upon creation of the cluster. With kubeadm you usually create your cluster like this:</p><pre>kubeadm init --config ./clusterconfig.yaml</pre><p>clusterconfig.yaml contains the configuration for your cluster. If you want to add the ExtendedResourceToleration simply add the following stub to your ClusterConfiguration object:</p><pre>---<br>apiVersion: kubeadm.k8s.io/v1beta2<br>kind: ClusterConfiguration<br>apiServer:<br>  extraArgs:<br>    enable-admission-plugins: ExtendedResourceToleration</pre><p>If you already have a cluster running, you will need to modify the kube-apiserver manifest on each master node running the Kubernetes API server. This manifest can be found at /etc/kubernetes/manifests/kube-apiserver.yaml Modify it so that the --enable-admission-plugins also contains ExtendedResourceToleration :</p><pre>apiVersion: v1<br>kind: Pod<br>metadata:<br>  ...<br>  name: kube-apiserver<br>  namespace: kube-system<br>spec:<br>  containers:<br>  - command:<br>    - kube-apiserver<br>    ...<br>    - --enable-admission-plugins=ExtendedResourceToleration</pre><p>Then restart kubelet with:</p><pre>systemctl restart kubelet</pre><p>And that’s it!</p><p>After kubectl apply-ing a GPU requesting pod, you can kubectl describe it and you’ll see the extra toleration:</p><pre>$ kubeclt describe pod ray-gpu-worker-pvvrs<br>Name:         ray-gpu-worker-pvvrs<br>...<br>Containers:<br>  ray-gpu-worker:<br>...<br>    Limits:<br>      cpu:                5<br>      memory:             40Gi<br>      nvidia.com/gpu:     1<br>...<br>Node-Selectors:  &lt;none&gt;<br>Tolerations:     node.kubernetes.io/not-ready:NoExecute op=Exists<br>                 node.kubernetes.io/unreachable:NoExecute op=Exists<br>                 nvidia.com/gpu:NoSchedule op=Exists</pre><p>Enjoy your fully automated toleration-adding-magic!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c180f5c5e76d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/robovision-ai/use-kubernetes-extendedresourcetoleration-to-keep-out-non-gpu-pods-c180f5c5e76d">Use Kubernetes’ ExtendedResourceToleration to keep out non-GPU pods</a> was originally published in <a href="https://medium.com/robovision-ai">Robovision AI</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Fixing Star Trek (Picard)]]></title>
            <link>https://samdecrock.medium.com/picard-season-2-please-give-use-hope-5c1b71124224?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/5c1b71124224</guid>
            <category><![CDATA[hope]]></category>
            <category><![CDATA[star-trek-picard]]></category>
            <category><![CDATA[future]]></category>
            <category><![CDATA[star-trek]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Sun, 29 Mar 2020 13:14:03 GMT</pubDate>
            <atom:updated>2024-09-05T19:46:12.752Z</atom:updated>
            <content:encoded><![CDATA[<p>Season 1 of <em>Star Trek: Picard</em> just wrapped, and as a lifelong fan, I’m left with mixed emotions. On one hand, I couldn’t help but feel goosebumps every time a familiar face from <em>The Next Generation</em> appeared on screen. But on the other, the show’s dark, brooding atmosphere left me feeling somewhat disconnected from the optimism that Star Trek once represented. At a time like this, especially in the midst of a global pandemic, we need stories that inspire hope, not ones that mirror the bleakness of our current reality.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G61d3uM_m1P7TU7eyx6iZg.jpeg" /><figcaption>Starfleet Headquarters in a bright San Francisco</figcaption></figure><p>One of the things I’ve always loved about Star Trek is how it portrayed a hopeful future. Starfleet was a symbol of what humanity could become if we kept striving for good — an intelligent, united society working together for the greater good. <em>Star Trek</em> has always been a beacon of hope, and that’s what the world needs now more than ever.</p><p>But <em>Picard</em>? It felt different, more dystopian than utopian. And that’s a shame. Our world already has enough darkness; we turn to Star Trek to remind us of the light.</p><h4>How Can We Fix Star Trek: Picard?</h4><p>Here are some suggestions to restore <em>Star Trek</em>’s hopeful vision:</p><h3>Ditch the cliffhangers</h3><p>One of the major problems with modern serialized storytelling is that it leaves the audience in a constant state of suspense. While suspense can be thrilling, it doesn’t offer the same kind of emotional satisfaction that earlier Star Trek episodes did. The beauty of pre-2005 <em>Star Trek</em> was in its episodic structure. Each episode was like a mini-movie with its own contained storylines — usually a blend of scientific intrigue and human connection. By the end of each episode, you felt a sense of closure and completeness.</p><p>The newer trend of multi-episode arcs is less rewarding. Instead of ending on a hopeful or thought-provoking note, we’re often left wondering what happens next, which diminishes the sense of optimism we crave. <em>Picard</em> could benefit from returning to a more episodic format, where each installment feels like a complete journey with its own challenges and resolutions.</p><h3>Leave the Vices in the Past</h3><p><em>Star Trek</em> has always been about imagining a better future, one where humanity has evolved past its current flaws. But in <em>Picard</em>, there’s an excess of smoking, drinking, and cursing, making it feel too grounded in today’s world. In a future where humanity has advanced so far, wouldn’t we have left behind habits like smoking and excessive drinking?</p><p>The beauty of <em>Star Trek</em> lies in its optimism. It shows us a world where humans have become more enlightened, where we’ve transcended some of our more destructive tendencies. To truly feel like a future worth aspiring to, <em>Star Trek</em> should reflect a society that has grown past these vices.</p><h3>Bring Back the Classical Score</h3><p>Music is one of the most powerful tools a show has to create atmosphere, and the music in <em>Picard</em> feels… off. The original <em>Star Trek</em> series and <em>The Next Generation</em> were full of sweeping orchestral scores that made you feel like you were embarking on a grand, hopeful adventure. Those stirring violins, brass, and woodwinds contributed to the sense that we were watching something truly uplifting.</p><p>In contrast, the modern soundtrack feels more subdued, relying heavily on electronic elements. While these can work in the right context, they don’t evoke the same sense of wonder and optimism that the old scores did. We need those soaring, triumphant melodies back, the ones that remind us of the hope that Star Trek was built upon.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UTjEvvAegyPTYhNqAb1tgA.png" /><figcaption>RSO Vienna’s performance of classic Star Trek themes perfectly captures that uplifting spirit. <a href="https://www.youtube.com/watch?v=YJ7zOFn0QYQ&amp;t=44">Listen here</a>.</figcaption></figure><h3>Shaping the Future Through Storytelling</h3><p>Maybe I’m being a little dramatic, but the stories we tell — especially those as iconic as <em>Star Trek</em> — have a profound impact on how people perceive the world and what they aspire to. Our kids aren’t just watching TV; they’re learning from it, absorbing the messages these shows put out. As creators, you have the power to influence how we think about the future.</p><p>And that future should be bright.</p><p><em>Thank you<br>A 35-year-old fan who grew up watching The Next Generation with his dad and hopes to one day watch a hopeful Star Trek with his own kids.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5c1b71124224" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to resize a gcePersistentDisk on GKE]]></title>
            <link>https://samdecrock.medium.com/how-to-resize-a-gcepersistentdisk-on-gke-60e60edaf89?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/60e60edaf89</guid>
            <category><![CDATA[kubernetes]]></category>
            <category><![CDATA[google-kubernetes-engine]]></category>
            <category><![CDATA[persistent-volume]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Wed, 11 Mar 2020 15:34:48 GMT</pubDate>
            <atom:updated>2020-03-11T15:40:59.778Z</atom:updated>
            <content:encoded><![CDATA[<p>If you’re like me and you defined your Google Persistent disk on GKE like this:</p><pre>volumes:<br>  - name: your-volume-name<br>    gcePersistentDisk:<br>      pdName: your-disk<br>      fsType: ext4</pre><p>Then, like me, you won’t have a partition table on this disk. Apparently it is possible to simply write to a disk without having a partition table.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WnWYWnhysKCuJAxprEHORg.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@matti_johnson?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Matti Johnson</a> on <a href="https://unsplash.com/s/photos/storage-increase?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>To increase a persistent disk defined in this way, start by going to your Disks and increase its size:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/451/1*CLhV1KTAVN16y6NTimdtoQ.png" /><figcaption>I’m increasing the size to 80GB here</figcaption></figure><p>Next, hit <strong><em>Save</em></strong> and click the link <strong><em>In use by</em></strong>. This will take you to the VM where the disk is attached to.</p><p>Next, click the <strong>SSH</strong> button to start an SSH session with the virtual machine:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/302/1*wi4MmXJDw8zqCo1FSN5YFg.png" /></figure><p>On the VM, you should be able to see your attached disk with:</p><pre>$ sudo df -h | grep /dev/sdb<br><strong>/dev/sdb   </strong>40G   27G   13G  69% /var/lib/kubelet/plugins/kubernetes.io/gce-pd/mounts/your-disk</pre><p>As you can see, it is still 40GB instead of the resized 80GB.</p><p>Next, resize it with</p><pre>$ sudo resize2fs /dev/sdb</pre><p>Verify its size:</p><pre>$ sudo df -h | grep /dev/sdb<br><strong>/dev/sdb   </strong>79G   27G   52G  35% /var/lib/kubelet/plugins/kubernetes.io/gce-pd/mounts/your-disk</pre><p>That’s it! Your Kubernetes cluster should now be able to use the extra disk space.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=60e60edaf89" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building your custom Raspbian image]]></title>
            <link>https://samdecrock.medium.com/building-your-custom-raspbian-image-8b54a24f814e?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/8b54a24f814e</guid>
            <category><![CDATA[cloning]]></category>
            <category><![CDATA[raspberry-pi]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Sun, 19 May 2019 10:41:53 GMT</pubDate>
            <atom:updated>2019-12-09T20:07:04.375Z</atom:updated>
            <content:encoded><![CDATA[<p>I recently had to prepare a <a href="https://www.amazon.com/CanaKit-Raspberry-Pi-Starter-Kit/dp/B07V4G63M1/ref=as_li_ss_tl?ie=UTF8&amp;linkCode=ll1&amp;tag=samdecrock-20&amp;linkId=a82267466f2d788271f9f54e3f895705&amp;language=en_US">Raspberry Pi</a> for a client, but it would take too long to send the Pi itself. I had to come with a way to prepare the Pi locally, create an image and send this image electronically.</p><p>The problem with creating an image of an SD card, is that it’s size is the size of the SD card. So if you take an image of 32 GB SD card, it has to be flashed back to a 32 GB SD card. That’s something you want to avoid.</p><p>In this blog post, I will explain how to prepare a Raspbian image on the Pi and turn it into a distributable image using Mac OS.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jl8_HojvgOXvl02xV7eesw.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/photos/YzUkt-X8TK4?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Hector Bermudez</a> on <a href="https://unsplash.com/search/photos/raspberry-pi?utm_source=unsplash&amp;utm_medium=referral&amp;utm_content=creditCopyText">Unsplash</a></figcaption></figure><p>First, start by removing the following line from the /boot/cmdline.txt file:</p><pre>init=/usr/lib/raspi-config/init_resize.sh</pre><p>This will prevent the Pi from expanding its file system upon boot. We need this removed, because we want to keep our partition as small as possible.</p><p>Next, boot from the SD card as you normally would. Install all the applications you want.</p><p>When your done, put your SD card back in to your Mac. Next, put the part your just removed from /boot/cmdline.txt back (or restore a backup if you took one).</p><p>Next, open the Terminal application and list your drives with</p><pre>diskutil list</pre><p>You will probably recognise your SD card as /dev/disk2:</p><pre>/dev/disk2 (internal, physical):<br>   #:                       TYPE NAME    SIZE       IDENTIFIER<br>   0:     FDisk_partition_scheme        *128.0 GB   disk2<br>   1:             Windows_FAT_32 boot    45.0 MB    disk2s1<br>   2:                      Linux         1.8 GB     disk2s2</pre><p>Find the block size with diskutil info /dev/disk2:</p><pre>$ diskutil info disk2<br>Device Identifier:        disk2<br>Device Node:              /dev/disk2<br>Whole:                    Yes<br>Part of Whole:            disk2<br>Device / Media Name:      SD Card Reader</pre><pre>Volume Name:              Not applicable (no file system)<br>Mounted:                  Not applicable (no file system)<br>File System:              None</pre><pre>Content (IOContent):      FDisk_partition_scheme<br>OS Can Be Installed:      No<br>Media Type:               Generic<br>Protocol:                 USB<br>SMART Status:             Not Supported</pre><pre>Disk Size:                128.0 GB (128043712512 Bytes) (exactly 250085376 512-Byte-Units)<br><strong>Device Block Size:        512 Bytes</strong></pre><pre>Read-Only Media:          No<br>Read-Only Volume:         Not applicable (no file system)</pre><pre>Device Location:          Internal<br>Removable Media:          Removable<br>Media Removal:            Software-Activated</pre><pre>Virtual:                  No</pre><p>Next, find the used size of the SD card’s partitions with sudo fdisk /dev/disk2:</p><pre>$ sudo fdisk /dev/disk2<br>Disk: /dev/disk2  geometry: 15567/255/63 [250085376 sectors]<br>Signature: 0xAA55<br>         Starting       Ending<br> #: id  cyl  hd sec -  cyl  hd sec [start -       size]<br>-------------------------------------------------------------------<br> 1: 0C    0 130   3 -    5 249  31 [ 8192 -      87851] Win95 FAT32L<br> 2: 83    6  30  25 -  219  68  41 [98304 -    3424256] Linux files*<br> 3: 00    0   0   0 -    0   0   0 [    0 -          0] unused      <br> 4: 00    0   0   0 -    0   0   0 [    0 -          0] unused</pre><p>I our case, the last partition starts at block 98304 and is 3424256 blocks big. That means the last used block of our SD card is 3522560 (98304+3424256).</p><p>To copy only the used part of our SD card to a file, use dd:</p><pre>sudo dd if=/dev/disk2 of=image.img bs=512 count=3522561</pre><p>Note that we added 1 to the last used block to be on the safe side. Set bs to the block size you found earlier.</p><p>You should now have a small image (image.img) which you can further zip:</p><pre>zip -r image.zip image.img</pre><p>In my case, this resulted in a file of 483 MB, similar to the size of original Raspbian image.</p><p>That’s it. Now you can easily flash the image to any SD card larger than 483 MB (which equals 99% of the SD cards in the world I guess).</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8b54a24f814e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cracking Mifare Classic NFC cards using the hardnested attack]]></title>
            <description><![CDATA[<div class="medium-feed-item"><p class="medium-feed-image"><a href="https://samdecrock.medium.com/cracking-mifare-classic-nfc-cards-using-the-hardnested-attack-506aab3ea305?source=rss-8f8336fa1565------2"><img src="https://cdn-images-1.medium.com/max/730/1*KKDAo0W2m7Ls8_KvjAQmeQ.jpeg" width="730"></a></p><p class="medium-feed-snippet">I&#x2019;ve always been a fan of hacking. Not to harm others but just for the fun of it.</p><p class="medium-feed-link"><a href="https://samdecrock.medium.com/cracking-mifare-classic-nfc-cards-using-the-hardnested-attack-506aab3ea305?source=rss-8f8336fa1565------2">Continue reading on Medium »</a></p></div>]]></description>
            <link>https://samdecrock.medium.com/cracking-mifare-classic-nfc-cards-using-the-hardnested-attack-506aab3ea305?source=rss-8f8336fa1565------2</link>
            <guid isPermaLink="false">https://medium.com/p/506aab3ea305</guid>
            <category><![CDATA[nfc]]></category>
            <category><![CDATA[hacking]]></category>
            <category><![CDATA[kali]]></category>
            <category><![CDATA[rfid]]></category>
            <dc:creator><![CDATA[Sam Decrock]]></dc:creator>
            <pubDate>Sat, 11 May 2019 13:30:58 GMT</pubDate>
            <atom:updated>2022-02-18T12:46:34.339Z</atom:updated>
        </item>
    </channel>
</rss>