<?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 Roman Dushko on Medium]]></title>
        <description><![CDATA[Stories by Roman Dushko on Medium]]></description>
        <link>https://medium.com/@rodush?source=rss-b8d093adaab0------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*q0IoHAO4HJ8ELlrpIjtkwA.jpeg</url>
            <title>Stories by Roman Dushko on Medium</title>
            <link>https://medium.com/@rodush?source=rss-b8d093adaab0------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 10:49:20 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@rodush/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[How to create a useful observability dashboard]]></title>
            <link>https://medium.com/@rodush/how-to-create-a-useful-observability-dashboard-76974bddf90b?source=rss-b8d093adaab0------2</link>
            <guid isPermaLink="false">https://medium.com/p/76974bddf90b</guid>
            <category><![CDATA[observability]]></category>
            <category><![CDATA[monitoring]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Roman Dushko]]></dc:creator>
            <pubDate>Tue, 25 Aug 2026 06:31:01 GMT</pubDate>
            <atom:updated>2026-09-17T09:50:45.396Z</atom:updated>
            <content:encoded><![CDATA[<p>In this post I would like to share my vision and experience of creation of useful operational dashboards. I will be talking here about the dashboards that software engineering teams are creating to monitor and support applications running in production environment.</p><p>Overall, the rule of thumb is this — track metrics which allow you to action on them, for instance to create an automated alert or make a decision that something needs investigation ASAP.<br>Separate Business Intelligence (BI) insights and metrics from Operational (Ops) insights.</p><h3>Core dashboard design principles</h3><ol><li>Put most important information top-left, and move in rows, following the <a href="https://www.nngroup.com/articles/f-shaped-pattern-reading-web-content/">F-Shape design principle</a>. Put less critical information down the page.<br>As we usually read information top-left → bottom-right since our childhood, our eyes and brains are just trained to digest information that way.<br><em>This mostly applies to cultures with the left-to-right (LTR) reading patterns!</em></li><li>Avoid too long pages, so scrollbar does not appear (say, on 24&quot; screen).<br>Stick to one screen to render most critical insights first. This is especially important for when you want your dashboard to be displayed on the big screen, where there will be noone to scroll the page down.</li><li>If observability platform supports it, use dashboard pagination. For instance, NewRelic supports <a href="https://docs.newrelic.com/docs/query-your-data/explore-query-data/dashboards/manage-your-dashboard/#add-pages">Pages</a>, Datadog calls it <a href="https://docs.datadoghq.com/dashboards/configure/#tabs">Tabs</a>. Tabs or Pages are extremely handy — they allow you to group information into logical units, and separate primary from secondary insights.</li><li>Do not mix operational insights with the business metrics!<br>It’s a very often mistake that engineers make when Product Owners or Product Managers use the tool which is designed for SRE-kind of monitoring to track business metrics for the sake of analytics. These are 2 different concerns, and both require proper attention and focus.<br>If you still find yourself in a situation that it is easier to use software-observability-oriented platform instead of the business intelligence oriented one (Looker, Tableau, MS’ Power BI), then at least make sure you have a dedicated dashboard for the product analytics data, or at worst — separate page on your operational dashboard.</li><li>Even though it might be useful to track operational metrics for success, try to focus on measuring of what’s going wrong (warnings and errors). In the end, you eventually need the monitoring software to tell you that something is not right, so you can quickly identify the problem and address it.<br>I am talking about metrics like number of requests. It’s fun to see your platform copes with the huge traffic, but you don’t really immediately act from this metric alone. You look at other signals — connection pool exhaustion, CPU, memory, etc.</li></ol><h3>What to track?</h3><p>It is not possible to give one best recipe obviously, as it all depends on what kind of application, software, or platform you are monitoring.</p><p>I will share my insights around what I usually put on the dashboard when I monitor backend API application. Some critical piece of information which is a great starting point to drill deeper.</p><ol><li><strong>API response error codes</strong>. This can be a bar chart or line.<br>My rule of thumb — <em>ideally</em> we don’t want to see 500 HTTP status codes at all, as it’s an Internal Server Error, so something 100% is not correct, and this needs to be looked into in the nearest time.<br>If you return 500 error to handle controlled errors (because you think it’s a problem originating from your server), reconsider the status code you use. Maybe it’s 401, or 403, or 429?</li><li><strong>Latency / response time</strong>. If it consistently goes high, something is not right, and you would want to investigate where is the culprit.<br>If you have external integrations (99.9% chance you do), it is well possible that the poor performance of your API is caused by poor performance of the upstream service.</li><li><strong>Async messages consumption lag</strong>. If you have asynchronous communication bridges (queues like SQS, Kafka), then it’s absolutely crucial to monitor consumption (lag in Kafka)!<br>In order to jump on a problem fast and reduce the potential impact — in case when the problem goes unnoticed for days, weeks, or even months — you want to see when messages are piling up unprocessed. Most probably, there is a problem with the consumer, and it’s not monitored correctly — nobody noticed the problem there.<br>If you have a Platform / SRE team who owns Kafka clusters, they can set up generic monitoring on all topics to be notified about the lag increasing over time.</li><li><strong>CPU and memory usage</strong> of lambda functions (AWS), Docker container runners (Fargate, ECS, Kubernetes), EC2 machines.<br>For containerized applications, monitor memory utilization together with the configured limits and application-specific behavior. Consistently high utilization with insufficient headroom can lead to OOM kills or undesirable GC behavior.</li><li><strong>DB resources utilisation</strong>: number of connections, CPU and memory usage of the cluster and replicas (if you have this setup), or a signle DB instance. Do not overcomplicate! Only visualise in the primary page of dashboard what helps you see and understand the problem.</li><li><strong>Behaviour of the upstream (external) dependencies </strong>— response times and status codes of responses. It’s important to quickly pin-point the problematic provider, so you can escalate and reach out to support team of the relevant platform or service.<br><strong>Track number of retries</strong> (assuming here you integrate with upstream depencies using the retry and backoff mechanism; if you don’t do it yet, consider adding for the sake of resilience and stability of your own service). This can help you identify consistently poor performer, so you can reach out to troubleshoot together with the engineers from the upstream service.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WMi_aVW9yhW__F8KPlJEBQ.png" /><figcaption>Example Dashboard layout. Generated with ChatGPT based on aforementioned information.</figcaption></figure><p>It makes sense to group information into pages like this:</p><p>Page 1 — Incident Triage: One screen. Highest-value signals. Almost no scrolling.<br>Page 2 — Infrastructure / dependencies: More detailed investigation.<br>Page 3 — Busines / Product insights: Optional metrics which might be needed to ensure business operations continuity.</p><p>If you are an advanced engineering / SRE team, you might want to focus your dashboards around SLI and SLO. High level signals (SLI) can give answer about reliability and performance (SLO) and to answer whether the customer expectations are met.</p><h3>What NOT to track?</h3><p>Information which is just a noise, rather than a useful, actionable signal.</p><p>Monitoring actionable information is useful because you can configure thresholds for warning and critical levels. When certain threshold is reached, the automated notification can be sent to the on-call paging tool, e.g. PagerDuty (P1, P2 incidents), as well as your daily communication media — Slack, MS Teams (P3, P4 incidents), or even email. I suggest to create dedicated channel for Operational Alerts and Notifications.</p><p>I have seen many dashboards where widgets are added simply “because we can”, and “why not”? Usually information rendered there does not help to troubleshoot an incident. If you like to see fancy widgets with little-to-no action, at least hide them “below the edge” or put them on a secondary / tertiary tab.</p><h3>Final words</h3><p>Having a condensed, refined, and well-thought dashboard helps engineering teams have a good grip on the state of health of their applications and services that run in Production environments.</p><p>Dashboards alone though are not enough for low MTTR and quick reaction.</p><p>Dashboards are passive sources of information, which require someone to proactively look at them to spot the signal — indicator of an upcoming disaster. For successful and reliable Observability you have to have automated alerting system, so you are notified (paged) when something goes off the track.</p><p>Don’t build a dashboard because you have metrics. Build it around the questions an engineer needs answered when production is unhealthy.</p><p>Metrics alone are not enough to tell why something is not working as expected. You also need to have robust logging system, where logged information can be correlated with the externally facing errors.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=76974bddf90b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Engineering managers should last longer]]></title>
            <link>https://medium.com/@rodush/engineering-managers-should-last-longer-81500261cd57?source=rss-b8d093adaab0------2</link>
            <guid isPermaLink="false">https://medium.com/p/81500261cd57</guid>
            <category><![CDATA[engineering-mangement]]></category>
            <category><![CDATA[management]]></category>
            <dc:creator><![CDATA[Roman Dushko]]></dc:creator>
            <pubDate>Mon, 13 Feb 2023 09:08:52 GMT</pubDate>
            <atom:updated>2023-02-13T09:11:34.087Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*r0P-YNfUiIsUebxk" /><figcaption>Photo by <a href="https://unsplash.com/@brianericksonco?utm_source=medium&amp;utm_medium=referral">Brian Erickson</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>I haven’t experienced the EM role yet, so it is hard for me to judge and justify from that stand point. My observations so far are solely from the engineer point of view, or “being managed” contributor.</p><p>It is always worrying to find out that your manager is leaving. First reaction is — damn, did they learn something bad about the organisation? (BTW, very often that’s true! Since managers are exposed to the information “first-hand”, they can also make the decision earlier). Even if that is not the case, and they decided to quit because they don’t see a growing opportunity for them in the organisation (which is also a bad sign, right?), the impact on their subordinates is huge. I am not exaggerating, it’s really huge!</p><p>Think about it. For several months, maybe a year, or even few years (if you are lucky enough to stay in the same company and have your manager around that long as well), you are building rapport with your manager. You build your profile, which is — although might be backed with some objective data — still <em>subjectively</em> perceived by your direct manager. Only the person who is close to you and in the same context, understand why certain things have happened the way they have happened.</p><p>When your direct manager leaves, the impact of that is pretty much similar to changing jobs. The only difference is — you don’t have a chance to negotiate your title and salary. And that is a very bad difference. To some extent, you carrier is pushed back, as you have to start with the new manager pretty much from scratch.</p><p>Also, the “blast radius” when the manager is leaving is times more than when an individual contributor / engineer is leaving. In a healthy work environment, a line manager efficiently manages 8–10 people. So, when the manager leaves, 8–10 people suddenly face the issue of being pushed backed in their progression.</p><p>I feel like the whole role of an engineering manager prescribes more responsibility. When you are an engineer and leave the code and project behind, well, there are few more fellas who were in the same boat as you, so they can still pedal without you (assuming, you’ve done your duty and handed over the knowledge properly). When you are are manager, you leave connections and people behind. You are a leader, right? The price of changing the leader is not comparable to the price of changing the “minion”.</p><p>I believe that a good manager, like a good captain of a ship, should stay on board as long as it takes, and truly be there for their people.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=81500261cd57" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Where is my Jira ticket? How to manage releases with Github Actions and Jira.]]></title>
            <link>https://medium.com/@rodush/where-is-my-jira-ticket-how-to-manage-releases-with-github-actions-and-jira-6e2ecbb53edd?source=rss-b8d093adaab0------2</link>
            <guid isPermaLink="false">https://medium.com/p/6e2ecbb53edd</guid>
            <category><![CDATA[jira]]></category>
            <category><![CDATA[github-actions]]></category>
            <category><![CDATA[project-management]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[automation]]></category>
            <dc:creator><![CDATA[Roman Dushko]]></dc:creator>
            <pubDate>Sun, 29 Jan 2023 22:48:16 GMT</pubDate>
            <atom:updated>2023-02-02T15:23:25.583Z</atom:updated>
            <content:encoded><![CDATA[<p>The story is about how it is possible to track where the Jira ticket is — whether it is in Stage, or in Production, using Github Actions and Jira Releases.</p><p><em>This article assumes the </em><a href="https://github.com/features/actions"><em>Github Actions</em></a><em> is used as the CI/CD tool.</em></p><h3>Introduction</h3><p>Many organisations follow Scrum/Agile practices these days. One of the most popular project and issue tracking systems being used out there is Atlassian’s Jira.</p><p>Jira allows to setup 2 types of the boards — Scrum and Kanban. Out-of-the-box, boards come with 3 typical columns — To Do, In Progress, Done. Very often people are not happy with the default setup and they start customising the boards and the flows to fit the vision and the purpose.</p><p>Let’s look at the typical lifecycle of a typical development “story” in a “Scrum” setup.</p><p>First, the item comes into the <a href="https://www.scrum.org/resources/what-is-a-product-backlog">Product backlog</a>. Once refined and ready to be picked, it will make its path to the <a href="https://www.scrum.org/resources/what-is-a-sprint-backlog">Sprint backlog</a>. Developer picks the task from the “To Do” column, assigning it to themself and dragging it to the “In Progress” column.</p><h3>The problem</h3><p>Now, the development work is “Done”, but that ticket is not ready to be shipped to production yet — none of the team mates reviewed the work, nor the work has been fully integrated into the main branch. In which status should the ticket be at this stage? The challenging part begins…</p><p>Let’s say, developer created a pull request in Github, and they wait for a review. Would be very nice to reflect the status of Jira ticket on the board, isn’t it? It’s not really “in progress” from the development perspective. The “In Review” column seems to be a natural fit to the flow.</p><p>So, we now have 4 columns — “To Do”, “In Progress”, “In Review”, “Done”:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*i1FIyHg9S-nbqGDZWm0qHg.png" /><figcaption>4 columns layout in a Scrum board</figcaption></figure><p>Nice. More transparency!</p><p>Okay, the ticket has been reviewed, and approved, and even merged into the ‘main’ branch. What next? Is the work “Done”? Well, unless your team has the full-fledged CI/CD (where “D” stands for ‘deployment’) environment, the ticket might at best be just ‘Integrated’ (I).</p><p>Once the story is integrated, perhaps, you want tag the snapshot and create the release candidate. Perhaps, in this snapshot not only one ticket, but several others — merged earlier — will be included.</p><h3>Solution</h3><p>Github UI allows to <a href="https://docs.github.com/en/repositories/releasing-projects-on-github/managing-releases-in-a-repository">manage releases</a> in the repository. The “Release” allows to create a new git tag, or choose an already existing one.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*soUonkM9lYMyYBC0EiP8ww.png" /><figcaption>Creating Release in Github UI</figcaption></figure><p>At this point, we can bring some automation to the stage to help us shepherd the herd of the tickets.</p><p>When “Release” is created in Github, we can synchronise it with the “Releases” in Jira. Using Github Actions it is possible to create Release in Jira:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JfI1g2pRK2gCZ2e4tyaH4g.png" /><figcaption>Jira Releases UI</figcaption></figure><p>It is also possible to find all the relevant Jira tickets which are included in the current tag, and mark them automatically in Jira with the relevant tag name (“Fix Versions” field in Jira).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/868/1*wfYac7QFig7sSn57eDDzaw.png" /><figcaption>Fix versions field in Jira ticket</figcaption></figure><p>For this magic to work, we need to introduce and follow some conventions. Mainly, the git commit summary has to include Jira issue name (like, ‘XYZ-123’, where XYZ is the project label). To enforce such format in NodeJS ecosystem, the <a href="https://www.npmjs.com/package/commitlint-config-jira">https://www.npmjs.com/package/commitlint-config-jira</a> can be used.</p><p>For each found ticket, the Github Action would then go to Jira via the REST API and update “Fix Versions” field for each relevant ticket.</p><p>With the “Fix versions” field been used, we essentially introduced another virtual state, w/o cluttering the board with the extra columns. Also, it’s very easy to see which tickets are part of the release from the Jira’s “Releases” UI.</p><p>When the actual deployment to production has happened, it is possible to find the relevant Release and Jira and mark is as “Released” with the actual date of deployment. This can be done automatically, using another Github Action (I don’t have one, still plan to create). It also can be done manually of course. The only thing, your Jira user has to have the “admin” right for the project to be able to release the Release.</p><p>To take care about automatic creation of the relevant Release in Jira, and to update Jira tickets with the new “fix version”, I’ve created the Github Action — <a href="https://github.com/marketplace/actions/sync-github-and-jira-releases">Sync Github and Jira Releases</a> . The action is published to the Github Actions Marketplace.</p><h3>Conclusion</h3><p>Using Github Releases, Github Actions, and Jira Releases, it is possible to much easier track the state of Jira ticket — whether it is released or not, and if released, part of which release it is.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6e2ecbb53edd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Siemens Dishwasher Error 15 Fix]]></title>
            <link>https://medium.com/@rodush/siemens-dishwasher-error-15-fix-19a2276d280c?source=rss-b8d093adaab0------2</link>
            <guid isPermaLink="false">https://medium.com/p/19a2276d280c</guid>
            <dc:creator><![CDATA[Roman Dushko]]></dc:creator>
            <pubDate>Sun, 16 Jan 2022 17:35:23 GMT</pubDate>
            <atom:updated>2022-01-16T18:05:59.812Z</atom:updated>
            <content:encoded><![CDATA[<h3>Siemens Dishwasher Error 15</h3><p>For my work I am a software engineer, for my spare time I am a DIY dude, and when something is broken at home I roll up my sleeves and go fix it.</p><p>Frankly, I am more successful at fixing mechanical things and devices, rather electronics — with that I mostly fail in my attempts… (only <a href="https://medium.com/@rodush">one thing</a> appeared to be a success).</p><p>This story is about the dishwasher Siemens SN59M531NL/74 (as appears, Siemens is actually a sibling of Bosch) and its Error 15.</p><p>One day I started the dishwasher in a mode which I use regularly. After some time my dishwasher started showing “Err15” on the display and producing sounds like it is trying to drain the water: the pomp motor was periodically starting, running for several seconds, then pausing for a while, and repeating it in the loop. This could last for hours. None of the buttons worked, not even the “Reset” or “Power” buttons. Then only way to interrupt this state was hard unplugging the machine from the socket. When plugged in back, it kept showing Error 15 on its LCD and kept doing those start-stop-start-stop draining attempts.</p><p>I started searching online for the error and possible ways for how to fix / reset the error. Some websites and articles which I found were mentioning that this error mean there is a leak in the system. I didn’t observe any water on the floor under the machine (I have the model embedded into the kitchen furniture).<br>It was also mentioning the potential failure of the automatic valve. The advice I found, was to check the filter in that valve. I did that — disconnected the valve from the supplying tap, checked the little filter, all seemed to be good. Plugging everything back and starting the machine again didn’t help — same Error15 on the display, same sound.</p><p>Some other article said that it could be that the water got onto the sensor and made it stuck in the “error” position. The advice was to tilt the machine approx. 45 degrees front facing, and make sure all the water came out from the machine, then let it dry, and try again.<br>Well, tilting the machine wasn’t a good option for me, unless I wanted to disassemble part of the kitchen. What I did though was I opened the front lid and let the machine dry for 1 day, while unplugged from the electrical socket. After that, I plugged the machine back and… the error was gone! I was very happy and started the regular washing program (Eco mode, 50 degrees). Everything look fine, until suddenly, after a few rinse cycles, the Error 15 appeared again on the display and that loop of pomp start-stop-start-stop started happening again. I got very upset, as I was out of potentials solutions.</p><p>The last option in my sleeve was to try disassemble the machine, and try figure out what could that be the reason for the damn Error 15.</p><p>It worth mentioning that there is pretty much no information about the dishwasher model which I have (mine is quite old, from 2001). Luckily, I managed to find Spanish website with some pictures about how to disable the machine to get to the section where the anti leak sensor is.</p><p>Once I removed the decorating MDF cover, it appeared to be quite obvious and simple what to do next. I unscrewed 4 screws and removed the front lid. There I found that sensor, which appeared to be a gauge floating in a foam plastic. And indeed, there I found water! I had no clue where did it came from… I grabbed a sponge, some old clothing and soaked the water around the foam plastic. Assembled everything back, plugged the machine into the socket, pressed the power button, and saw no more Err15! Success!</p><p>Then, I started the dishwashing program. After ~20 minutes of running, the Err15 appeared again on the display, and the machine stuck in the known draining attempt loop.</p><p>Same procedure — unplugged from the electricity socket, disassembled the front part again, this time more confident, knowing what I am doing. And there was the water again, and the gauge was floating, pressing the sensor lever… Damn!</p><p>Again, soaked the water out, let the machine dry for a short while, and decided to give it a try without putting the leads back.<br>I started the program and started waiting. After some time, the machine started its first rinsing cycle, where the pomp started draining the water and I heard that new fresh water is taken from the supply. I grabbed the pocket light and started looking into the machine. I noticed the water slowly started to appear under the float! After some examination around, I realised that the water is coming from the joint of the main chamber and the part where all that motor / pomp assembly resides.</p><p>My first idea was — the sealing had failed… I started looking online for the sealing ring to buy, managed to find some for the similar Bosch / Siemens model.</p><p>But then I noticed that there are 2 screws under the drain cover, inside the dishwasher chamber (where the coarse parts filter is). I try to screw them and felt like they were quite loose.</p><p>After tightening the screws, I started the washing cycle again. Waited for the rinse cycle, heard the fresh washer is coming, and started looking for the traces of water under the chamber. This time everything was dry, and no water was coming from under the joint. Success! I waited for the program to finish, made sure the leak sensor is dry and there is no water under the float. Certainly a Success!</p><p>I assembled everything back. The machine works just fine, for almost a half year already!</p><p>Now, the fun stuff. Here are some pictures and a link to the video to help anyone else with the similar issue.</p><figure><img alt="Screws which hold the pomp and filter assembly connected with the dishwasher machine chamber" src="https://cdn-images-1.medium.com/max/1024/1*nMRguKshNqexq8LFOgIzLw.jpeg" /><figcaption>The screws which are holding the pomp assembly connected to the chamber</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*q7RxxVJPue1CocytQLzkRQ.jpeg" /><figcaption>The screws which are holding the decorative MDF panel</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wcaesf4EO-Uz19i6ju-0Rg.jpeg" /><figcaption>The bottom part of the machine. The bagging covers the stainless steel cover.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DoHY9gibvP_jK1zoblZCVA.jpeg" /><figcaption>The bolt (2x) holding the cover to the underneath chamber with motor, pomp, and electronics</figcaption></figure><figure><img alt="TX20 socket to remove the lid and tighten the bolts" src="https://cdn-images-1.medium.com/max/1024/1*PImI0WT6w6gDv4yf_BC_iw.jpeg" /><figcaption>Socket TX20 which is needed to disassemble the cover and tighten the bolts</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*l3dv98VHDKICmbgFx_pU0A.jpeg" /><figcaption>Anti-leak sensor. Floating plate made of the foam plastic or something like that.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Sv_m0eoZw-bxt671aSAbog.jpeg" /></figure><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F61kF_h98fSo%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D61kF_h98fSo&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F61kF_h98fSo%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/88ab650828684ca5562a4fe03ae00814/href">https://medium.com/media/88ab650828684ca5562a4fe03ae00814/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=19a2276d280c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Adopting the zsh. NodeJS developer’s story.]]></title>
            <link>https://medium.com/@rodush/adopting-the-zsh-nodejs-developers-story-9af28047460a?source=rss-b8d093adaab0------2</link>
            <guid isPermaLink="false">https://medium.com/p/9af28047460a</guid>
            <category><![CDATA[nodejs]]></category>
            <category><![CDATA[nvm]]></category>
            <category><![CDATA[n]]></category>
            <category><![CDATA[nvmrc]]></category>
            <category><![CDATA[zsh]]></category>
            <dc:creator><![CDATA[Roman Dushko]]></dc:creator>
            <pubDate>Sun, 05 Jan 2020 21:20:00 GMT</pubDate>
            <atom:updated>2020-01-06T17:25:07.305Z</atom:updated>
            <content:encoded><![CDATA[<h3>Managing node versions with n, nvmrc and zsh</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*K_TnEkCGcGbIZq7L" /><figcaption>Photo by <a href="https://unsplash.com/@markusspiske?utm_source=medium&amp;utm_medium=referral">Markus Spiske</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>Recently macOS Catalina (10.15.x) switched to <strong>zsh</strong> being the default shell environment. Before that it was just a normal bash (3.2.57).</p><p>To say the truth, I am not that much of a bash-scripting geek, and I don’t have many fancy adjustments, plugins and configurations added on top of the default setup — just the bare minimum needed for my daily work (some tiny helper one-liners, some environment variables, and that is mostly all).</p><p>My colleagues around were already using <strong>zsh</strong> with some fancy setup (mostly, <a href="https://github.com/ohmyzsh/ohmyzsh">oh-my-zsh</a>). I wasn’t bought into it and didn’t want to switch from my familiar environment, but since I was forced by macOS, I finally made a decision to install <strong>zsh </strong>as a default shell.</p><p>After installation of the <strong>oh-my-zsh </strong>(with a simple `<em>brew install oh-my-zh</em>` command), all my bash-related configuration was turned off (fortunately, the <strong>oh-my-zsh</strong> backs up the file). So, I manually had to merge the content of my previous .bash_profile and the new .zshrc files. I also had to disable few lines which were specific to <strong>bash</strong>.</p><p>It took me some time to enable back the <strong>git</strong> completion (I am a <em>console</em> git user and don’t use any graphical plugins or tools to do my daily job with <strong>git</strong>). Thankfully, I found <a href="https://medium.com/@oliverspryn/adding-git-completion-to-zsh-60f3b0e7ffbc">this great post</a> which helped me get things up and running quite quickly.</p><p>Currently, I work as a <em>NodeJS</em> programmer. In the company I currently work for, we have to deal with many micro services (NodeJS projects). We also need to deal with the mess around different <strong>node </strong>versions and constantly have to switch back and force as we switch among different micro-services (different git projects).</p><p>I like to try new things and quite often do things differently, compared to masses. As such, after having used nvm for some time, I wanted to give a try to another node-version-manager — n . It worked quite well, so I stayed with it until now.</p><p>Recently, I have heard from my teammate that he uses some fancy script which works together with nvm and automatically switches node ‘s environment to the <strong>node</strong> version specified in the .nvmrc file.<br>Some “<em>googling</em>” gave me a <a href="https://medium.com/@udeshi_82511/manage-multiple-nodejs-versions-47ea415f4f4d">result</a>, which explained how that <em>magic</em> works.<br>Since I am not using nvm , the script didn’t work for me with a simple copy-paste approach. As such, I decided to not look further, but to implement my own (maybe super-simplified, but pragmatic and working) version of the script specifically for n:</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/940f193adcc062226f1b4d03f6d05eed/href">https://medium.com/media/940f193adcc062226f1b4d03f6d05eed/href</a></iframe><p>Once added, restart a Terminal window or open a new tab of it, switch to any project which contains a .nvmrc file and observe n switching <em>automagically</em> to the exact required version of <strong>node</strong>. Hooray! No more manual operations!</p><p>Enough for now. Not sure if there will be much more to add. When I learn something interesting, I will share as another (linked) post.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9af28047460a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building real conversations with Dialogflow and App.Chatbot]]></title>
            <link>https://medium.com/dazn-tech/building-real-conversations-with-dialogflow-and-app-chatbot-4a10e5fa8c03?source=rss-b8d093adaab0------2</link>
            <guid isPermaLink="false">https://medium.com/p/4a10e5fa8c03</guid>
            <category><![CDATA[bots]]></category>
            <category><![CDATA[conversation-design]]></category>
            <category><![CDATA[dialogflow]]></category>
            <category><![CDATA[chatbot-design]]></category>
            <dc:creator><![CDATA[Roman Dushko]]></dc:creator>
            <pubDate>Wed, 11 Dec 2019 09:11:01 GMT</pubDate>
            <atom:updated>2019-12-11T09:11:01.279Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="robot made of parts from constructor for children" src="https://cdn-images-1.medium.com/max/1024/0*3O2njXZ30ouSEEol" /><figcaption>Photo by <a href="https://unsplash.com/@rocknrollmonkey?utm_source=medium&amp;utm_medium=referral">Rock&#39;n Roll Monkey</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p><strong>TL;DR</strong> Dialogflow’s UI is not the best to design more-or-less complex conversations. It has poor search capabilities across intents, as well as very limited functionality around managing “nested” dialogs (those which have in- and out- contexts).<br>Using only Dialogflow, it might be quite a “painful” experience. It makes sense to look into 3rd party conversation design tools which support API to extract the designed dialogs and import them into Dialogflow. One of such tools is App.Chatbot.</p><p>There are quite a lot of blog posts and publications in the Internet about how to build chat bots using Dialogflow platform. I am not going to repeat the basics here (terms like intents, entities, contexts, etc.).<br>Instead, I will share our experience in building the chat bot which supports quite a long conversation around any of supported topics.</p><p>Our Customer Services team decided to offload some work to the modern technology — AI, and with a small team of developers we started investigating how we can help there.</p><p>Based on some preliminary triaging, the choice of suitable platform to build the bot was made in favour of Dialogflow. Shortly, the main reasons to choose Dialogflow as a bot engine were — wide range of supported languages, reasonable price even for the Enterprise plan (Standard plan with 3 req/s is completely free!) and mature NLP support out of box.</p><p>.When we just started, as many other developers, we looked around and played with Dialogflow trying to understand its UI, concepts and specifics. We tried to build several simple conversations, using parameters, contexts and entities. Everything looked simple and very promising…</p><p>Then the time came to start building real, longer conversations. Company hired a conversation designer — nice guy, he also has decent experience in script-writing for <em>big name</em> companies.<br>His task was to create conversations for our chatbot. Those conversations are supposed to simulate the FAQ and troubleshooting steps, that a live support person usually does: answering specific questions about cancellation or pausing the subscription, troubleshooting issues with sign-in and sign-up, authentication issues, or issues with playback, etc.</p><p>Our conversation designer started building the conversations. As appeared in practice, Dialogflow’s UI is not the best tool for designing (wire-framing) of dialogs for a chatbot. As such, another platform was picked for these needs — <a href="https://app.chatbot.com/">https://app.chatbot.com</a> <br>The web interface of App.Chatbot is quite an advanced one, and allows to prototype bots. Also it is possible to test the designed bot directly inside the platform, as well as to connect the bot to several popular social and messaging platforms (Facebook, Slack, and few others):</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8GEGvpyQIUWlfVeL3hjhow.png" /><figcaption>App.Chatbot interface to create Bots (Stories)</figcaption></figure><p>To help users get answers for their queries, our conversation designer decided to go on with an approach of “quick replies”, assuming users would click one of those offered buttons, instead of typing free text messages.</p><p>With such approach, conversation around each topic quickly turned into a nested structure (nodejs/javascript developers would recognise it as a “callback hell”):</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w2sbGSiTxSHQ_dz1beQ2pg.png" /><figcaption>“callback hell” of a nested conversation</figcaption></figure><p>Obviously, clicking the “quick reply” button should not be the only way to interact with our bot. Users should be able to also type the answer, with the similar words as it is scripted for the “quick reply” button. For instance, when Bot replied with a choice of 3 quick reply subjects — “My account”, “Payment Issues” and “Sign up”, and user types in the answer “<em>I have a problem with my payment</em>”, then the bot should be able to categorise the reply as “Payment issues” and define the context respectively.</p><p>Looking at the picture above and the requirement around the support for freely typed text, it may lead you to a conclusion that we heavily rely on contexts while building our bot conversations.</p><p>Now, as I mentioned earlier, Dialogflow’s UI is not the best when it comes to designing, or even configuring complex, especially contextually linked, nested conversations.</p><p>There are several ways how developers (bot builders) usually structure their dialogs (or conversations), using the means provided by the platform (Dialogflow).</p><p>Option #1, is to mark related conversations with some text label prefix, e.g.: [PAYMENTS].<br>For a slice of the conversation from the screenshot above, that might look something like:</p><blockquote>[MY ACCOUNT] Payment issues<br>[PAYMENTS] Payment payment<br>[PAYMENTS] Payment failed: sufficient funds<br>[PAYMENTS] Payment failed: countries match<br>…<br>[PAYMENTS] Payment failed: countries mismatch</blockquote><p>Each line above represents one intent (which has many utterances inside). Dialogflow keeps everything in a flat structure. You can search only by intent name, not by utterance (“<em>usersays</em>”)…</p><p>Option #2, is to use the follow-up intents, using the Dialogflow’s feature to build nested conversation — when <em>output</em> context of the previous intent matches the <em>input </em>context of the next intent.<br>Dialogflow developers did some work to visualise such dependencies. But the nested tree is collapsed by default:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GB1MMY1XfMMDIx6XDKeV4w.png" /><figcaption>By default, follow-up intents are collapsed</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zhfyZSRzTizdP-RIQdDFdQ.png" /><figcaption>Need to click the ^ icon to expand each level independently</figcaption></figure><p>From the screenshots above you may see that it is pretty much impossible to find any nested intent unless you remember its name at least approximately (then one could use the Intent Search bar):</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zgdVTRCuSyPZO1DQU1LvFQ.png" /><figcaption>“countries match” contained in the title of one of the nested intents</figcaption></figure><p>we would have to manually expand each and every intent and verify if search term matches part of the intent’s name (no highlight is built in, unfortunately). Not a scalable approach, honestly speaking.</p><p>In that sense, App.Chatbot is somewhat simpler to use, because we may click the “Expand all” (+) button and then use browser’s “Find” (Cmd+F) feature to search for a term and such term will be highlighted:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S19gbWlCDdO0Qh0h0JAc3A.png" /><figcaption>Using the “Expand all” and browsers “Find on page” feature to find needed intent</figcaption></figure><p>Since there is no out-of-the-box integration between App.Chatbot and Dialogflow, migrating manually all those intents and conversations would be super tedious, error prone and complicated. That would also require a ton of effort and attention from the people doing that job.</p><p>Fortunately, both App.Chatbot and Dialogflow have API which allows to export stories (from App.Chatbot), and programmatically import configured intents (into Dialogflow).</p><p>The only problem with API is of course, JSON schemas and concepts within each platform mismatch completely!.. <br>After a close examination of the JSON document which we received in a response to the API call, we understood that it is still feasible and does not require a lot of effort to convert data exported from App.Chatbot into a Dialogflow’s JSON document (separate for each Intent).</p><p>As a result, we managed to connect 2 systems together: design conversations using the App.Chatbot, and use NLP capabilities from Dialogflow.<br>We periodically execute script and migrate conversations from one system into another, and never manually touch Intents imported into Dialogflow. For us source of truth, so far, is App.Chatbot.<br>How we retrain the model by adding new user utterances will become a topic for another publication.</p><h3>Conclusion</h3><p>Dialogflow is a great platform to start building chatbots. With its simple interface and advanced NLP engine, it is a piece of cake to build your own agent. Having talked to several people I realised, that not only developers can build bots, but even marketers and customer support people!</p><p>At the same time, once you start adding more and more conversations, and those conversations become more complex and start having a nested structure (linked via input and output contexts), Dialogflow’s UI turns into a hard to use one — from the point of overviewing a complete conversation, but also from the point of searching for a specific intent to add further modifications.</p><p>It makes sense to look into other options and solutions, which have more user- friendly interface to design and configure conversations.<br>One of such solutions is to use App.Chatbot platform, and integrate it with Dialogflow via API.</p><p><em>Many thanks to </em><a href="https://medium.com/u/f59697bddf64"><em>Darko Antic</em></a><em> for his decent review and very nice suggestions.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4a10e5fa8c03" width="1" height="1" alt=""><hr><p><a href="https://medium.com/dazn-tech/building-real-conversations-with-dialogflow-and-app-chatbot-4a10e5fa8c03">Building real conversations with Dialogflow and App.Chatbot</a> was originally published in <a href="https://medium.com/dazn-tech">DAZN Engineering</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Diary, 2019]]></title>
            <link>https://medium.com/series/diary-2019-287fe22c2d19?source=rss-b8d093adaab0------2</link>
            <guid isPermaLink="false">https://medium.com/p/287fe22c2d19</guid>
            <dc:creator><![CDATA[Roman Dushko]]></dc:creator>
            <pubDate>Fri, 11 Jan 2019 10:29:09 GMT</pubDate>
            <atom:updated>2019-09-23T13:53:04.137Z</atom:updated>
            <content:encoded><![CDATA[<p>11 Jan, 2019</p><p>Today I figured out how to update `pinentry-curses` to `pinentry-mac`:</p><ol><li>update (or create one, if you don’t have it yet) `~/.gnupg/gpg-agent.conf` file by specifying which pinentry program to use:<br>&gt; pinentry-program /usr/local/bin/pinentry-mac</li><li>reload gpg-agent:<br>&gt; gpg-connect-agent reloadagent /bye</li><li>I also update simlink for pinentry (/usr/local/bin/pinentry) to point to the pinentry-mac binary <br>Although not sure if it was needed.</li></ol><p>22 Jan, 2019</p><p>Figured out the proper way to filter tagged features (Cucumber) when start wdio runner:</p><p>` yarn test:e2e:headless — cucumberOpts.tagExpression=’@search’`</p><p>18 Feb, 2019</p><p>Suddenly was unable to commit changes into my repository, getting these messages:</p><p>&gt; error: gpg failed to sign the data<br>&gt; fatal: failed to write commit object</p><p>It appeared my gpg key got expired a day before.</p><p>Decided to just extend expiration period with:</p><p>&gt; gpg — quick-set-expire {KEY_FINGERPRINT} {NEW_DATE}</p><p>To get the key’s fingerprint use:</p><p>&gt; gpg -k</p><p>(if you have one key, it is easier)</p><p>23 Sep, 2019</p><p>Trying to debug node app while running jest unit test.</p><p>Executing the command like this:</p><p>&gt; node — inspect ./node_modules/.bin/jest — runInBand ./test/component</p><p><em>— runInBand</em> is crucial, otherwise node inspector doesn’t stop on the debugging point! (chrome dev tools doesn’t connect to the debugger)</p><p>Kudos to: <a href="https://artsy.github.io/blog/2018/08/24/How-to-debug-jest-tests/">https://artsy.github.io/blog/2018/08/24/How-to-debug-jest-tests/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=287fe22c2d19" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Start blogger’s life]]></title>
            <link>https://medium.com/@rodush/start-blogger-s-life-90fecfcd5759?source=rss-b8d093adaab0------2</link>
            <guid isPermaLink="false">https://medium.com/p/90fecfcd5759</guid>
            <category><![CDATA[initiation]]></category>
            <dc:creator><![CDATA[Roman Dushko]]></dc:creator>
            <pubDate>Tue, 04 Aug 2015 10:07:56 GMT</pubDate>
            <atom:updated>2015-08-04T10:07:56.270Z</atom:updated>
            <content:encoded><![CDATA[<p>Decided to give a try to blogging about diverse sides of my life and everything related to it.</p><p>Actually the idea was to involve my wife to write posts. Because she actually able to write really good stories. The only thing she needs is a startpoint, a good example and a motivation.</p><p>That is for you, my lovely wife Oksana!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=90fecfcd5759" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>