<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom"><title>Song's blog</title><link href="https://blog.midu.com.au/" rel="alternate"></link><link href="https://blog.midu.com.au/feeds/all.atom.xml" rel="self"></link><id>https://blog.midu.com.au/</id><updated>2019-10-30T16:01:00+11:00</updated><entry><title>Survival guide when working behind company proxy - Docker</title><link href="https://blog.midu.com.au/posts/2019-10/survive-proxy-docker/" rel="alternate"></link><published>2019-10-30T16:01:00+11:00</published><updated>2019-10-30T16:01:00+11:00</updated><author><name>Song.Jin</name></author><id>tag:blog.midu.com.au,2019-10-30:/posts/2019-10/survive-proxy-docker/</id><summary type="html">&lt;p&gt;a guide to work with company proxy&lt;/p&gt;</summary><content type="html">&lt;h2&gt;the story&lt;/h2&gt;
&lt;p&gt;I was trying to login to AWS ECR from a machine inside company proxy. Out of
blue, I got time-out error!&lt;/p&gt;
&lt;div class="highlight"&gt;&lt;pre&gt;&lt;span&gt;&lt;/span&gt;Error response from daemon: Get https://aws-account.dkr.ecr.... request cancelled
while waiting for connection (Client.Timeout exceeded while waiting for headers)
&lt;/pre&gt;&lt;/div&gt;


&lt;h2&gt;the explore&lt;/h2&gt;
&lt;p&gt;From my experience, it is comming from company proxy which blocks your network
traffic quietly again. But problem is most of Google-search results suggest to
reconfig &lt;em&gt;systemd&lt;/em&gt; which is not available to my machine( it doesn't have it, and
I can't install it)&lt;/p&gt;
&lt;h2&gt;the solution&lt;/h2&gt;
&lt;p&gt;Let cut to the cheese. Assuming you have root permission to the machine,&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;add proxy settings to &lt;code&gt;/etc/sysconfig/docker&lt;/code&gt; file.&lt;/li&gt;
&lt;/ul&gt;
&lt;div class="highlight"&gt;&lt;pre&gt;&lt;span&gt;&lt;/span&gt;sudo su -
echo &amp;#39;export http_proxy=http://&amp;lt;company-proxy:proxy-port&amp;gt;&amp;#39; &amp;gt;&amp;gt; /etc/sysconfig/docker
echo &amp;#39;export https_proxy=http://&amp;lt;company-proxy:proxy-port&amp;gt;&amp;#39; &amp;gt;&amp;gt; /etc/sysconfig/docker
echo &amp;#39;export no_proxy=[list-of-internal-domains]&amp;#39; &amp;gt;&amp;gt; /etc/sysconfig/docker
&lt;/pre&gt;&lt;/div&gt;


&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;restart docker daemon
&lt;code&gt;sudo service docker restart&lt;/code&gt;&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;prove if you could run Docker login command correctly, response should be &lt;em&gt;Login succeed&lt;/em&gt;
&lt;code&gt;$(aws ecr get-login --no-include-email)&lt;/code&gt;&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;
&lt;h2&gt;the traps&lt;/h2&gt;
&lt;p&gt;I feel it is nice to share traps I fell into, so that you wouldn't experience
the same pain I have been through.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;no_proxy should contain company internal domains, and no more. Those domains
that are behind company firewall should be here, but anything else should go
through the forwardproxy.&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;It is good to align your tools with what are popular on the market. If I used
any AMIs that contains &lt;em&gt;systemd&lt;/em&gt;, I could have copy&amp;amp;paste solution from the first
Google search result for this error message. The AMI I used is Amazon 1, which
doesn't have &lt;em&gt;systemd&lt;/em&gt;. So if I do it again, I would just pick an AMI shipped with
&lt;em&gt;systemd&lt;/em&gt; which is popular on the Internet.&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;</content><category term="devops"></category></entry><entry><title>How to set correct resolution on Ubuntu 18.04 LTS</title><link href="https://blog.midu.com.au/posts/2019-03/ubuntu-resolution/" rel="alternate"></link><published>2019-03-22T22:35:00+11:00</published><updated>2019-03-22T22:35:00+11:00</updated><author><name>Song.Jin</name></author><id>tag:blog.midu.com.au,2019-03-22:/posts/2019-03/ubuntu-resolution/</id><summary type="html">&lt;p&gt;how I setup correct resolution for my Ubuntu&lt;/p&gt;</summary><content type="html">&lt;h1&gt;Why writing this&lt;/h1&gt;
&lt;p&gt;I want to run Ubuntu on my Lenovo laptop with a 2560x1440@60Hz monitor. But there is
no such resolution in Display setting's resolution list.&lt;/p&gt;
&lt;p&gt;And I went through many online articles. finally I found one method works for me&lt;/p&gt;
&lt;h1&gt;The code&lt;/h1&gt;
&lt;p&gt;modelline is the string I got after run &lt;code&gt;umc 2560 1440 55 --rbt&lt;/code&gt;&lt;/p&gt;
&lt;table class="highlighttable"&gt;&lt;tr&gt;&lt;td class="linenos"&gt;&lt;div class="linenodiv"&gt;&lt;pre&gt; 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13&lt;/pre&gt;&lt;/div&gt;&lt;/td&gt;&lt;td class="code"&gt;&lt;div class="highlight"&gt;&lt;pre&gt;&lt;span&gt;&lt;/span&gt;&lt;span class="ch"&gt;#!/usr/bin/env bash&lt;/span&gt;

&lt;span class="nv"&gt;outlet&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="k"&gt;$(&lt;/span&gt;xrandr &lt;span class="p"&gt;|&lt;/span&gt; grep &lt;span class="s1"&gt;&amp;#39;connected primary&amp;#39;&lt;/span&gt; &lt;span class="p"&gt;|&lt;/span&gt; awk &lt;span class="o"&gt;{&lt;/span&gt;&lt;span class="s1"&gt;&amp;#39;print $1&amp;#39;&lt;/span&gt;&lt;span class="o"&gt;}&lt;/span&gt; &lt;span class="k"&gt;)&lt;/span&gt;

&lt;span class="nv"&gt;modelline&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="s1"&gt;&amp;#39;&amp;quot;2560x1440x54.97&amp;quot; 221.000000 2560 2608 2640 2720  1440 1443 1447 1478  +HSync -VSync&amp;#39;&lt;/span&gt;

&lt;span class="nv"&gt;model&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="k"&gt;$(&lt;/span&gt;&lt;span class="nb"&gt;echo&lt;/span&gt; &lt;span class="nv"&gt;$modelline&lt;/span&gt; &lt;span class="p"&gt;|&lt;/span&gt; awk &lt;span class="o"&gt;{&lt;/span&gt;&lt;span class="s1"&gt;&amp;#39;print $1&amp;#39;&lt;/span&gt;&lt;span class="o"&gt;}&lt;/span&gt;&lt;span class="k"&gt;)&lt;/span&gt;

xrandr --newmode &lt;span class="nv"&gt;$modelline&lt;/span&gt;

xrandr --addmode &lt;span class="nv"&gt;$outlet&lt;/span&gt; &lt;span class="nv"&gt;$model&lt;/span&gt;

xrandr --output &lt;span class="nv"&gt;$outlet&lt;/span&gt; --mode &lt;span class="nv"&gt;$model&lt;/span&gt;
&lt;/pre&gt;&lt;/div&gt;
&lt;/td&gt;&lt;/tr&gt;&lt;/table&gt;

&lt;h1&gt;The trick I found the hard way&lt;/h1&gt;
&lt;ul&gt;
&lt;li&gt;
&lt;p&gt;&lt;code&gt;cvt&lt;/code&gt; command that everyone is talking, is not really working in my case, instead &lt;code&gt;umc&lt;/code&gt; works like a charm!&lt;/p&gt;
&lt;/li&gt;
&lt;li&gt;
&lt;p&gt;do not let frequency very high, or you would have blinky screen! I would tweak the frequency number such as the 55 in &lt;code&gt;umc 2560 1440 55 --rbt&lt;/code&gt;, so that I can see resulting frequency is 221MHz which ensures a stable display.&lt;/p&gt;
&lt;/li&gt;
&lt;/ul&gt;</content><category term="ubuntu resolution"></category></entry><entry><title>Kubernetes Apiserver latency alert with Prometheus</title><link href="https://blog.midu.com.au/posts/2018-09/apiserver-prom-alert/" rel="alternate"></link><published>2018-09-22T14:35:00+10:00</published><updated>2018-09-22T14:35:00+10:00</updated><author><name>Song.Jin</name></author><id>tag:blog.midu.com.au,2018-09-22:/posts/2018-09/apiserver-prom-alert/</id><summary type="html">&lt;p&gt;how I setup alert for kubernetes apiserver high latency&lt;/p&gt;</summary><content type="html">&lt;blockquote&gt;
&lt;p&gt;Good devops engineers won't let anyone else to tell them that your systems have problems.&lt;/p&gt;
&lt;/blockquote&gt;
&lt;p&gt;My team has realised it the hard way that the apiserver and the etcd module
behind it has limited resource to deal with objects inside Kubernetes. For
example, if one of your program trying to query large amount of objects, apiserver
and the etcd module will freak out. As a result, all apiserver users would
experience very long response time. Good devops engineers won't let anyone
else to tell them that your systems have problems.
That is why I am trying to add an alert for such events.&lt;/p&gt;
&lt;h2&gt;Problem break down&lt;/h2&gt;
&lt;p&gt;While it is true that there are delays from network component even the operators
laptop, we can only monitor objective metrics scaped from server. Please be aware
the trade-off here is that we are ignoring a part of high latency perceived by
users: non-server related delays.&lt;/p&gt;
&lt;p&gt;In Prometheus, there are 2 ways we can monitor. One is tracking
absulte value of the latency, the other is finding increase rate of it.&lt;/p&gt;
&lt;h2&gt;My 1st failed attempt&lt;/h2&gt;
&lt;p&gt;In my first attempt, I tried a metric &lt;code&gt;apiserver_request_latencies_summary&lt;/code&gt;.
Good thing about it is that the metric shows an absolute value in mirco seconds.
However, it is not very suitable for a responsive alerting system. The metric
summarises data from a sliding window upto 5 hours. So if your system has been
okey, then suddenly went bad. The high rate of latency won't be picked up until
very long time later that averaged value of the past 5 hours has crossed your
set threshold. The delay in alerting could be upto 1.5 hours, and alert won't
cool down until same delay time past actual end of the incident. Alert setting
in PromQL:
&lt;code&gt;avg_over_time(apiserver_request_latencies_summary{job="kubernetes-apiservers", verb!~"WATCH|CONNECT", quantile="0.9"}[5m]) / 1000 &amp;gt; 3000&lt;/code&gt;&lt;/p&gt;
&lt;h2&gt;My 2nd attempt&lt;/h2&gt;
&lt;p&gt;As an improvement to the above alert, I found a project on Github called
&lt;code&gt;Prometheus Operator&lt;/code&gt;. One of their &lt;em&gt;rules&lt;/em&gt; uses metric
 &lt;code&gt;apiserver_request_latencies_bucket&lt;/code&gt;, then visualise it with &lt;code&gt;histogram_quantile&lt;/code&gt;
The query makes the alert very responsive to sudden surge of apiserver latency.
&lt;code&gt;avg by(resource, verb) (histogram_quantile(0.95, rate(apiserver_request_latencies_bucket{verb!~"^(?:WATCH|CONNECT)$"}[5m])) / 1e+06 &amp;gt; 0) &amp;gt; 1&lt;/code&gt;&lt;/p&gt;
&lt;p&gt;In the mentioned &lt;em&gt;rules&lt;/em&gt;, thresholds were set to &amp;gt;1:warning, and &amp;gt;4:critical&lt;/p&gt;
&lt;h2&gt;Personal reflection&lt;/h2&gt;
&lt;p&gt;External factors make it harder for me to work on this task. First is lack of
proper documentation about the metrics. Not finding any offical documentation,
I had to dig down to source code, only to find a few inline code comments helping
make sence of the metrics. But when I came back to Prometheus, what? Metrics in
source code has different names here, and some metrics here can't be found in
source code.&lt;/p&gt;
&lt;p&gt;Another thought is that the high-latency scenario needs some effort to re-produce
in order to test your alert. For my case, my team came up with this code to create
huge amount of objects to query which will lead to high latency. You would need
to generate at least 7 thousand objects to execrise apiserver capability.&lt;/p&gt;
&lt;p&gt;Final thoughs are:
- why sudden surge of objects could happen
- could we have pagination at apiserver level or at etcd level?
- could performance optimisation help? such as, using high IOPS instance for etcd.&lt;/p&gt;
&lt;table class="highlighttable"&gt;&lt;tr&gt;&lt;td class="linenos"&gt;&lt;div class="linenodiv"&gt;&lt;pre&gt; 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12&lt;/pre&gt;&lt;/div&gt;&lt;/td&gt;&lt;td class="code"&gt;&lt;div class="highlight"&gt;&lt;pre&gt;&lt;span&gt;&lt;/span&gt;&lt;span class="ch"&gt;#!/usr/bin/env bash&lt;/span&gt;
 &lt;span class="nb"&gt;set&lt;/span&gt; -e
 &lt;span class="nv"&gt;counter&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="m"&gt;0&lt;/span&gt;
&lt;span class="k"&gt;while&lt;/span&gt; true&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="k"&gt;do&lt;/span&gt;
       &lt;span class="nv"&gt;counter&lt;/span&gt;&lt;span class="o"&gt;=&lt;/span&gt;&lt;span class="k"&gt;$((&lt;/span&gt;counter+1&lt;span class="k"&gt;))&lt;/span&gt;
       sleep &lt;span class="m"&gt;0&lt;/span&gt;.1
       cat config-map.yaml&lt;span class="p"&gt;|&lt;/span&gt; sed s/&lt;span class="se"&gt;\$&lt;/span&gt;INDEX/&lt;span class="nv"&gt;$RANDOM&lt;/span&gt;/g &lt;span class="p"&gt;|&lt;/span&gt; kubectl apply -f - &amp;gt; /dev/null &lt;span class="p"&gt;&amp;amp;&lt;/span&gt;
       &lt;span class="k"&gt;if&lt;/span&gt; &lt;span class="o"&gt;[[&lt;/span&gt; &lt;span class="s2"&gt;&amp;quot;&lt;/span&gt;&lt;span class="nv"&gt;$counter&lt;/span&gt;&lt;span class="s2"&gt;&amp;quot;&lt;/span&gt; -gt &lt;span class="m"&gt;30000&lt;/span&gt; &lt;span class="o"&gt;]]&lt;/span&gt;&lt;span class="p"&gt;;&lt;/span&gt; &lt;span class="k"&gt;then&lt;/span&gt;
         &lt;span class="nb"&gt;echo&lt;/span&gt; &lt;span class="s2"&gt;&amp;quot;Counter: &lt;/span&gt;&lt;span class="nv"&gt;$counter&lt;/span&gt;&lt;span class="s2"&gt; times reached; Exiting loop!&amp;quot;&lt;/span&gt;
         &lt;span class="nb"&gt;exit&lt;/span&gt; &lt;span class="m"&gt;1&lt;/span&gt;
       &lt;span class="k"&gt;fi&lt;/span&gt;
&lt;span class="k"&gt;done&lt;/span&gt;
&lt;/pre&gt;&lt;/div&gt;
&lt;/td&gt;&lt;/tr&gt;&lt;/table&gt;

&lt;p&gt;&lt;em&gt;config-map.yaml&lt;/em&gt;&lt;/p&gt;
&lt;div class="highlight"&gt;&lt;pre&gt;&lt;span&gt;&lt;/span&gt;---
apiVersion: v1
kind: ConfigMap
metadata:
  labels:
    app: prometheus
    component: server
    repo: ops-kube-system
    deployed-with: kt
    owner: songjin
  name: test-cm-$INDEX
  namespace: kube-system
&lt;/pre&gt;&lt;/div&gt;


&lt;h2&gt;Summary&lt;/h2&gt;
&lt;p&gt;Hopefully you can learn from my journey above, and save some struggle.&lt;/p&gt;</content><category term="devops"></category></entry><entry><title>My reading list</title><link href="https://blog.midu.com.au/posts/2018-06/read-list/" rel="alternate"></link><published>2018-06-04T21:28:00+10:00</published><updated>2018-06-04T21:28:00+10:00</updated><author><name>Song.Jin</name></author><id>tag:blog.midu.com.au,2018-06-04:/posts/2018-06/read-list/</id><summary type="html">&lt;p&gt;My reading list&lt;/p&gt;</summary><content type="html">&lt;h2&gt;general&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Apprenticeship Patterns&lt;/strong&gt; It is for anyone who want to improve their skills
and mindset which lead to those improvements. Not only for Juniors, for sure.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2&gt;programming&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;The Go programming language. I have been blown away by the good ideas in the book.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2&gt;kubernetes&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Kubernetes up and running&lt;/strong&gt; The book is great for new starters to get up to
speed of k8s administration&lt;/li&gt;
&lt;/ul&gt;</content><category term="devops"></category></entry><entry><title>CICD with PelicanBlog</title><link href="https://blog.midu.com.au/posts/2018-05/cicd-pelicanblog/" rel="alternate"></link><published>2018-05-23T00:00:00+10:00</published><updated>2018-05-23T00:00:00+10:00</updated><author><name>Song.Jin</name></author><id>tag:blog.midu.com.au,2018-05-23:/posts/2018-05/cicd-pelicanblog/</id><summary type="html">&lt;p&gt;how I build this blog&lt;/p&gt;</summary><content type="html">&lt;p&gt;Back in time, I knew PelicanBlog was a Python module that can generate static
website for me, as well as GithubPages was a good place to host the static website
generated.&lt;/p&gt;
&lt;p&gt;But the pain was I had to compile and update content manually, every and each time.&lt;/p&gt;
&lt;p&gt;This time, I decide to automate this bit!&lt;/p&gt;
&lt;h2&gt;how it is setup&lt;/h2&gt;
&lt;p&gt;Key components of this blog include:&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;a Github repo containing &lt;a href="https://github.com/SongGithub/songgithub.github.io-src"&gt;source code&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;another Github repo containing &lt;a href="https://github.com/SongGithub/songgithub.github.io"&gt;compiled static web files&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;a CICD &lt;a href="https://travis-ci.org/SongGithub/songgithub.github.io-src"&gt;pipeline&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;a &lt;a href="https://blog.getpelican.com/"&gt;static site generator&lt;/a&gt;&lt;/li&gt;
&lt;li&gt;a domain provider to get you custom URL, instead of generic site URL like foo.github.io&lt;/li&gt;
&lt;/ul&gt;
&lt;h2&gt;challenges&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;how the CI agent get permission to write code to one of my repos ?&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;I considered solutions such as SSH keys, Github personal token. SSH key didn't
get to my final thoughts because it is complicated to setup. You need to generate
a pair of keys, embed the Pub key in the repo, then ensure your CI agent has got
the private key and added to current ssh-agent. And repeat yourself this process
for other repos. Even though Github personal access token may have too much power
over my other repos that are not relevant to the project, I still think it is
overall better solution. Because it is simple, thus less error prone.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;given CI has got permission to write back to repos, how to carry out the task ?&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;It has been my trick that one can easily do stuff with permissions originated
from the Github personal access token, by simply ensure &lt;em&gt;origin's&lt;/em&gt; url is using
HTTPS, and have the token embeded like so
&lt;code&gt;https://&amp;lt;your-token&amp;gt;@github.com/&amp;lt;repo-owner&amp;gt;/&amp;lt;repo-name&amp;gt;.git&lt;/code&gt;
It can't be simpler than that!&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;how to have a faster feedback loop&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;A valid fast feedback loop should be its environment as similar as possible to
the one that is running on the server. One great tool is Docker which helps to
isolate your dependencies, and make it more repeatable on the server, and less
surprise that &lt;em&gt;"but it works on my laptop, why?"&lt;/em&gt;&lt;/p&gt;
&lt;p&gt;Another point will be having source code separated from configuration as much as
possible. So when you see someone put those configurable variables into source code,
you would realise it has this kind of weakness, hence it is hard to debug.&lt;/p&gt;
&lt;p&gt;Lastly, for this particular project, local dev server makes getting feedbacks quickly.
The &lt;strong&gt;Pelican&lt;/strong&gt; can compile my source code in Markdown files and stylesheets the same
way as it would do on the server. So I can tweak the website as frequent as
several times a minute.&lt;/p&gt;
&lt;ul&gt;
&lt;li&gt;how to have a robust build solution, not a flaky solution?&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;That was one of my lesson here. I read an article online that uses Github
submodule embeded in source code. I learned it the hardway that if you use something
in way that it is not suppose to do, be aware of hidden gotchas. The submodule
in this case, make it very compicated to run simple &lt;code&gt;git push/checkout&lt;/code&gt;. Because it
is made to &lt;em&gt;read&lt;/em&gt; code from a commit.&lt;/p&gt;</content><category term="devops"></category></entry><entry><title>1 yr anniversary with Kubernetes</title><link href="https://blog.midu.com.au/posts/2018-05/1-yr-k8s/" rel="alternate"></link><published>2018-05-22T16:01:00+10:00</published><updated>2018-05-22T16:01:00+10:00</updated><author><name>Song.Jin</name></author><id>tag:blog.midu.com.au,2018-05-22:/posts/2018-05/1-yr-k8s/</id><summary type="html">&lt;p&gt;1 yr anniversary with k8s&lt;/p&gt;</summary><content type="html">&lt;p&gt;I joined a company that builds its platform with Kubernetes last year this time.
I have since amazed how much this technology can help Developers and the Company!&lt;/p&gt;
&lt;h2&gt;Its benefits:&lt;/h2&gt;
&lt;h3&gt;It saves time from developers to deploy&lt;/h3&gt;
&lt;p&gt;Developers do not need to worry about infrastructure running their workloads, as well
as alerting and monitoring. All they need to know now, is how to write simple
Kubernetes config yaml files with help from our team.&lt;/p&gt;
&lt;h3&gt;It improves software development mentality&lt;/h3&gt;
&lt;ul&gt;
&lt;li&gt;Immutable infrastruture&lt;/li&gt;
&lt;li&gt;Immutable workload.&lt;/li&gt;
&lt;li&gt;Docker images as artifacts&lt;/li&gt;
&lt;/ul&gt;
&lt;p&gt;Before, developers liked to maintain a long running machine which is special. Time
spent on manually configure them was an investment. So when something fails, the
person on call need to jump into the Server, spending time to resolve the problem.&lt;/p&gt;
&lt;p&gt;Now, they don't need to and they can't do that anymore with Dockerised app running
on Kubernetes cluster.&lt;/p&gt;
&lt;h2&gt;I have improved myself since then&lt;/h2&gt;
&lt;p&gt;There are people skills, tech skills, as well as table tennis skills gained in
 a large enterprise in Australia.&lt;/p&gt;
&lt;p&gt;From technical perspective, I try to reduce my gap with some top-of-range
develoeprs I work with. In particular, is about to find root cause of a problem
with existing skills and knowledge.&lt;/p&gt;
&lt;p&gt;From sports perspective, I do enjoy plays and a sense of craftmanship in
table tennis skills. That is, never stop to refine your skills in an area.
It is called Kaizen by Toyota people.&lt;/p&gt;</content><category term="devops"></category></entry></feed>