August 03, 2021

While working on a side project yesterday, I was reading through the Pod manifests for some services. And one of the fields, I read there was qosClass: BestEffort.

I might be very ignorant before (as I never paid attention to this), but I wanted to understand what this QoS class (or Quality of Service class) meant.

So, yea below is what I understood & it is a very easy concept. 🙂

So, when we create pods in a kubernetes cluster, kubernetes assigns one of following QoS classes to (every) Pod:

• Guaranteed
• Burstable
• BestEffort

Each of these QoS class listed above have something to do with whether the Pod objects have properly defined resources (CPU, memory) or not.

Guaranteed

A Pod (including all the init containers and app containers) will get a QoS class of Guaranteed, if:

apiVersion: v1
kind: Pod
metadata:
  name: test-pod
  namespace: default
spec:
  containers:
  - name: test-container
    image: nginx
    resources:
      limits:
        memory: "50Mi"
        cpu: "200m"
      requests:
        memory: "50Mi"
        cpu: "200m"

• Every Container in the Pod must have a memory limit and a memory request set.

• For every Container in the Pod, the memory limit must be equal to the memory request.

  For example ~

  • limits(memory) = requests(memory) = 50 Mi
• Every Container in the Pod must have a CPU limit and a CPU request.

• For every Container in the Pod, the CPU limit must equal the CPU request.

  For example ~

  • limits(cpu) = requests(cpu) = 200m

Burstable

A Pod will get a QoS class of Burstable, if:

apiVersion: v1
kind: Pod
metadata:
  name: test-pod
  namespace: default
spec:
  containers:
  - name: test-container
    image: nginx
    resources:
      limits:
        memory: "50Mi"
      requests:
        memory: "100Mi"

• The Pod does not meet the criteria for QoS class Guaranteed.

• At least one Container in the Pod has a memory or CPU request.

  For example ~

  • requests(memory) > limits(memory)

BestEffort

A Pod will get a QoS class of BestEffort, if:

apiVersion: v1
kind: Pod
metadata:
  name: test-pod
  namespace: default
spec:
  containers:
  - name: test-container
    image: nginx

• The containers in the Pod must not have any memory or CPU limits or requests.

I wanted to upgrade one of my servers from CentOS 7 to CentOS 8. The following are the steps I followed :

Check the kernel version

Before we start the upgrade process, let us see the kernel version.

# uname -a
Linux centostest 3.10.0-1127.el7.x86_64 #1 SMP Tue Mar 31 23:36:51 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

Installing basic packages

Now let us begin with installing the basic packages.

yum install epel-release -y

EPEL provides a set of additional packages for CentOS, RHEL from the Fedora sources.

yum install yum-utils rpmconf

yum-utils is a collection of utility tools and programs for managing yum repositories, installing different packages such as debug packages, source packages.

rpmconf searches for .rpmnew, .rpmsave, and .rpmorigfiles and asks the user what to do with them.

The following cammand will check the configuration files of all packages and ask users for input.

rpmconf -a

Install dnf package manager

dnf is the replacement for yum in the newer version of the OS.

yum install dnf

Remove yum

Remove yum, the default package manager for CentOS 7, and yum configuration file to avoid any conflict with dnf. Since for CentOS 8 dnf is the primary package manager.

dnf -y remove yum yum-metadata-parser

rm -Rf /etc/yum

Getting CentOS 8 release packages

We are all set to upgrade from CentOS 7 to CentOS 8, but we need to upgrade the system before that.

dnf upgrade

This will update packages to their latest versions that are both available and resolvable.
I have installed CentOS 8 release packages and also for EPEL.

dnf install http://mirror.centos.org/centos/8/BaseOS/x86_64/os/Packages/{centos-linux-repos-8-2.el8.noarch.rpm,centos-linux-release-8.4-1.2105.el8.noarch.rpm,centos-gpg-keys-8-2.el8.noarch.rpm}
dnf -y upgrade https://dl.fedoraproject.org/pub/epel/epel-release-latest-8.noarch.rpm
dnf clean all

Removed the old kernel core of CentOS 7 and conflicting packages along with it.

rpm -e `rpm -q kernel`
rpm -e --nodeps sysvinit-tools

CentOS 8 system upgrade

The following packages, dracut-network and rpmconf, conflicts upgrade process, therefore removed them.

dnf remove dracut-network rpmconf

With the following dnf command, we ask to download all the non deltarpms for CentOS 8.

dnf -y --releasever=8 --allowerasing --setopt=deltarpm=false distro-sync

Installing a new kernel

To get the new kernel for CentOS 8, run the following command.

dnf -y install kernel-core

The final step would be to install CenOS 8 minimal packages.

dnf -y groupupdate "Core" "Minimal Install"

After all done:

# cat /etc/os-release
NAME="CentOS Linux"
VERSION="8"
ID="centos"
ID_LIKE="rhel fedora"
VERSION_ID="8"
PLATFORM_ID="platform:el8"
PRETTY_NAME="CentOS Linux 8"
ANSI_COLOR="0;31"
CPE_NAME="cpe:/o:centos:centos:8"
HOME_URL="https://centos.org/"
BUG_REPORT_URL="https://bugs.centos.org/"
CENTOS_MANTISBT_PROJECT="CentOS-8"
CENTOS_MANTISBT_PROJECT_VERSION="8"

Now run reboot to complete the upgrade.

Log back to the server, check the kernel version, and TaDa! We have CentOS 8 running on the system.

# uname -a
Linux centostest 4.18.0-305.10.2.el8_4.x86_64 #1 SMP Tue Jul 20 17:25:16 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux

Happy upgrading :)

July 26, 2021

Last friday, I attended a kubernetes #in-dev community session around the Cluster API (CAPI) project.

This was my very first time knowing what CAPI project is about (although, I’ve come across the name quite a lot of times before). And well, I found it to be a very interesting one for me. The reason being CAPI is a project, a tooling that helps with managing cluster infrastructure as declarative configuration. And this is the area that I’m concerned with, on a regular basis.

This blog is just me taking notes as I try to understand the concepts of the CAPI project & further play, around while seting it up locally.

Before I start with the commands, here is the definition for the CAPI (from he Cluster API Book).

Cluster API is a Kubernetes sub-project focused on providing declarative APIs and tooling to simplify provisioning, upgrading, and operating multiple Kubernetes clusters.

Started by the Kubernetes Special Interest Group (SIG) Cluster Lifecycle, the Cluster API project uses Kubernetes-style APIs and patterns to automate cluster lifecycle management for platform operators. The supporting infrastructure, like virtual machines, networks, load balancers, and VPCs, as well as the Kubernetes cluster configuration are all defined in the same way that application developers operate deploying and managing their workloads. This enables consistent and repeatable cluster deployments across a wide variety of infrastructure environments.

With that, let’s get started now!

It is required to have the following installed & setup on the machine (as pre-requisites):

• kubectl
• Kind
• Docker

STEP-1: Once the above pre-requisites are setup, the first step is, creating a local kubernetes cluster locally. (I’ll do this with kind)

• Later in the post, I’ll be using Docker infrastructure provider (while playing around with CAPI to generate workload cluster, something we will discuss later in the post)

  Think of infrastructure provider as any of these cloud providers like AWS, GCP, Azure, etc that provides compute and resources, required in order to spin up a cluster. Because I are going to do it locally, I’ll be using docker containers as my infrastructure provider (i.e. use docker containers as computer resources) to create a cluster using CAPI.

  I require to create a custom kind config file as the following (for allowing the Docker provider to access Docker on the host):

  cat > kind-cluster-with-extramounts.yaml <<EOF
  kind: Cluster
  apiVersion: kind.x-k8s.io/v1alpha4
  nodes:
  - role: control-plane
    extraMounts:
      - hostPath: /var/run/docker.sock
        containerPath: /var/run/docker.sock
  EOF
  

• And then I create a kind cluster, with the above config file.

  kind create cluster --config kind-cluster-with-extramounts.yaml

STEP-2: Installing the clusterctl CLI tool. It handles the lifecycle of a Cluster API management cluster.

• Since, I’m on a linux OS, following are the commands I’m using:

  $ curl -L https://github.com/kubernetes-sigs/cluster-api/releases/download/v0.4.0/clusterctl-linux-amd64 -o clusterctl
  $ chmod +x ./clusterctl
  $ sudo mv ./clusterctl /usr/local/bin/clusterctl
  $ clusterctl version
  

STEP-3: At this step, I’ll use clusterctl tool to transform the (one I spinned up above) kind cluster into a management cluster by using clusterctl init. The command accepts as input a list of providers to install.

A Management cluster is a Kubernetes cluster that manages the lifecycle of Workload Clusters. A Management Cluster is also where one or more Infrastructure Providers run, and where resources such as Machines are stored.

• Since, I’m using Docker infrastructure provider, I will add the --infrastructure docker flag to my command.

  clusterctl init --infrastructure docker

  

• The above command actually created the following CRD (custom resource definitions) & CR (custom resources) in the kind cluster (specific to the infrastructure provider I mentioned above)

  

STEP-4: Now, next step is to create a Workload Cluster.

A workload cluster is a cluster created by a ClusterAPI controller, which is not a bootstrap cluster, and is meant to be used by end-users, as opposed to by CAPI tooling.

• clusterctl generate cluster with input flags is used to generate a YAML template for creating a workload cluster. In my case, I’m using docker as infrasture provider, I’ll use the following command:

  clusterctl generate cluster test-workload-cluster --flavor development \
  --kubernetes-version v1.21.2 \
  --control-plane-machine-count=3 \
  --worker-machine-count=3 \
  > test-workload-cluster.yaml
  

  This command will populate the test-workload-cluster.yaml file with a predefined list of Cluster API objects; Cluster, Machines, Machine Deployments, etc. required to create a workload cluster with the (⬆️) listed cluster control & worker machine configs.

• Apply the test-workload-cluster.yaml file to create the workload cluster.

  kubectl apply -f test-workload-cluster.yaml

  

STEP-5: Now is the time to verify the workload cluster deployment & actually accessing it.

• Check the status of the cluster

  kubectl get cluster

• Use clusterctl to view of the cluster and its resources

  clusterctl describe cluster test-workload-cluster

• Check the status of the control plane

  kubectl get kubeadmcontrolplane

  Note: *The control plane won’t be Ready until I’ll insall a CNI (container network interface) in the following step.

STEP-6: Deploy a CNI solution.

• First, retrieve the workload cluster kubeconfig:

  clusterctl get kubeconfig test-workload-cluster > test-workload-cluster.kubeconfig

• Here, it will use calico for an example.

  kubectl --kubeconfig=./test-workload-cluster.kubeconfig apply -f https://docs.projectcalico.org/v3.18/manifests/calico.yaml
  

• After a short while, the nodes should be up & running

  kubectl --kubeconfig=./test-workload-cluster.kubeconfig get nodes
  

  Note that in the above commands, we are using the workerload cluster kubeconfig, and so, with that we are accessing the workload cluster, not the kind (management cluster) we spinned up for the first.

STEP-7: Finally, step is to delete the workload cluster & the management cluster

• Delete the workload cluster

  kubectl delete cluster test-workload-cluster

• Delete the management cluster

  kind delete cluster

References

• The Cluster API Book - Quick Start
• Overview of clusterctl
• Cluster API Concepts

Last night I was in a panel along with Juan Andrés Guerrero-Saade organized by Aveek Sen, the topic was "Tips on how journalists can avoid getting snooped". You can watch the recording at Youtube.

But this post is not about that. It is about Zoom. Just before logging into the call, I made sure that the name is changed while joining the call, generally my daughter uses the Zoom and her name was mentioned before. I personally have almost zero zoom usage (except 2-3 times in last 1 year). But, after logging into the call, zoom again went back to the older name, and did not allow me to change it during the session. I kept trying during the session without any luck. I don't know why did they do this or why I could not find a way to change my name, but I feel this is really stupid.

Over time, I have learnt a lot from the open-source software community, not just in India but from various parts of the world. I have the privilege of learning from the best software engineers in this world. A lot of great mentors have influenced me and helped me in shaping my way into OSS. These interactions were valuable to me, being a self-taught software engineer.

While talking to one junior from my alma mater, I realized there is still a gap between people wanting to get mentored and good mentors in the OSS community. There are outstanding programs like DGPLUG Summer Training which help new contributors learn how to survive in the open-source community by covering in breadth the mechanics. Still, people usually don’t discover on their own. I understand how frustrating it is to get stuck when you are exploring something. Having a mentor/partner with whom you can discuss doubts is vital in that situation.

As a way of giving back to the community that I have learned so much from, I am pledging 4 hours per week of my time to talk with folks who seek mentorship in OSS, Career journey, or, in general, talk about anything common interests.

Just go to https://calendly.com/palnabarun/1-on-1 to schedule a slot.

curl is most probably the highest used software in the world. I generally use it daily (directly) in the various scripts at the SecureDrop, starting from inside of Dockerfiles, to Ansible roles or in CI. I never read much about various options available other than a few very basic ones. So I decided to look more into the available options. Here are a few interesting points from that reading:

Protocols supported

This is a very long list, I had no clue!!!!

• dict
• file
• ftp
• ftps
• gopher
• gophers
• http
• https
• imap
• imaps
• ldap
• ldaps
• mqtt
• pop3
• pop3s
• rtsp
• scp
• sftp
• smb
• smbs
• smtp
• smtps
• telnet
• tftp

The --version flag will tell you this details and many things more.

$ curl --version
curl 7.76.1 (x86_64-redhat-linux-gnu) libcurl/7.76.1 OpenSSL/1.1.1k-fips zlib/1.2.11 brotli/1.0.9 libidn2/2.3.1 libpsl/0.21.1 (+libidn2/2.3.0) libssh/0.9.5/openssl/zlib nghttp2/1.43.0
Release-Date: 2021-04-14
Protocols: dict file ftp ftps gopher gophers http https imap imaps ldap ldaps mqtt pop3 pop3s rtsp scp sftp smb smbs smtp smtps telnet tftp 
Features: alt-svc AsynchDNS brotli GSS-API HTTP2 HTTPS-proxy IDN IPv6 Kerberos Largefile libz Metalink NTLM NTLM_WB PSL SPNEGO SSL TLS-SRP UnixSockets

User agent string

We can use -A to provide any given string as User Agent.

$ curl -A "browser/kd 1.0.0" http://httpbin.org/get
{
  "args": {}, 
  "headers": {
    "Accept": "*/*", 
    "Host": "httpbin.org", 
    "User-Agent": "browser/kd 1.0.0", 
    "X-Amzn-Trace-Id": "Root=1-60ec8985-05acbf60769f18855db925f1"
  }, 
  "origin": "xxx.xxx.xxx.xxx", 
  "url": "http://httpbin.org/get"
}

Download via sequence

We can specify the range in our command. The first example downloads different index files from my blog.

curl -O "https://kushaldas.in/index-[1-10].html"

A few more examples:

curl -O "http://example.com/[1-100].png"

curl -O "http://example.com/[001-100].png"

curl -O "http://example.com/[0-100:2].png"

curl -O "http://example.com/section[a-z].html"

curl -O "http://example.com/{one,two,three,alpha,beta}.html"

Parallel downloads

You can pass -Z to the above command to download in parallel. Try it out yourself.

Configuration file

We can create a configuration file to hold all the command line parameters and pass it to the curl command via -K flag.

curl -K configs.txt https://kushaldas.in

One flag each line, comments via #.

# Change user agent
--user-agent "ACAB/1.0"

Curl checks for a default configuration file at ~/.curlrc unless called with -q.

To view only the headers

-I to check the headers returned by the server.

$ curl -I https://kushaldas.in
HTTP/2 200 
server: nginx/1.14.2
date: Sun, 18 Jul 2021 06:40:37 GMT
content-type: text/html; charset=utf-8
content-length: 25977
last-modified: Mon, 05 Jul 2021 15:24:32 GMT
etag: "60e32430-6579"
strict-transport-security: max-age=31536000
onion-location: https://kushal76uaid62oup5774umh654scnu5dwzh4u2534qxhcbi4wbab3ad.onion
permissions-policy: interest-cohort=()
x-frame-options: DENY
x-content-type-options: nosniff
referrer-policy: strict-origin
accept-ranges: bytes

DNS over HTTPS

We can force curl to use a DoH server for the DNS query, pass --doh-url flag to the command along with the server address.

curl -I --doh-url https://adblock.doh.mullvad.net/dns-query https://securedrop.org

I have been maintaining dgplug infrastructure for sometime now. We have a blogging platform, blogs.dgplug.org for our community. We were planning to migrate it to the latest version of Operating System and the blogging software, but my post covid symptoms hindered that. A few days ago I finally managed to do that.

The server was running Bionic and the last release of writefreely. The plan was to migrate :

• Bionic to Focal (via do-release-upgrade)
• Upgrade writefreely to v0.13.1

While I was at it, I added a few HTTP response headers to the nginx configuration. Though the CSP (Content Security Headers) are yet to be added.

Now off to reading the actual blogposts at https://blogs.dgplug.org/read :)

While running your Ansible playbook, we can have the option of passing the argument from the command line. With the help of this, we define a generic playbook and use command line arguments to control the play. Let see how

Let's first start with the playbook, where we have defined the variable x is the playbook.

---
- hosts: all
  vars:
    x: 45
  tasks:
    - debug: var=x

In this playbook, we will pass the value of x from the command line arguments with the help -e “value”.

---
- hosts: all
  tasks:
    - debug: var=x

>> ansible-playbook playbook_nmae.yml -e "x=56"

Now let's write a playbook which can uninstall/install a package with the help of command line arguments.

---
- name: Install/Unistall pacakges with command line
  hosts: all
  became: 'yes'
  tasks:
    - name: 'working with {{pkg}}'
      apt:
        name: '{{pkg}}'
        state: '{{req_state}}'

Now, from the command line, we can pass the package name and state of the package.

>>ansible-playbook playbook_name.yml -e  "pkg=nginx req_state=present"

>>ansible-playbook playbook_name.yml -e  "pkg=nginx req_state=absent"

Cheers!

#100DaysToOffload #Ansible

I've been playing around with containers for a few years now. I find them very useful. If you host your own, like I do, you probably write a lot of nginx configurations, maybe apache.

If that's the case, then you have your own solution to get certificates. I'm also assuming that you are using let's encrypt with certbot or something.

Well, I didn't want to anymore. It was time to consolidate. Here comes Traefik.

---
version: '2.3'

services:

traefik:
  container_name: traefik
  image: "traefik:latest"
  restart: unless-stopped
  mem_limit: 40m
  mem_reservation: 25m

volumes:
  - "/var/run/docker.sock:/var/run/docker.sock:ro"

command:
  - --entrypoints.web.address=:80
  - --entrypoints.websecure.address=:443

ports:
  - "80:80"
  - "443:443"

- --entrypoints.web.http.redirections.entryPoint.to=websecure
- --entrypoints.web.http.redirections.entryPoint.scheme=https

- --providers.docker

- --log.level=INFO

- --api.dashboard=true

- --metrics.prometheus=true

- --certificatesresolvers.cloudflareresolver.acme.email=<your@email.here>
- --certificatesresolvers.cloudflareresolver.acme.dnschallenge.provider=cloudflare
- --certificatesresolvers.cloudflareresolver.acme.storage=./acme.json

- "./traefik/acme.json:/acme.json"

environment:
  - CLOUDFLARE_EMAIL=<your-cloudflare@email.here>
  - CLOUDFLARE_API_KEY=<your-api-key-goes-here>

labels:
  - "traefik.http.routers.dashboard-api.rule=Host(`dashboard.your-host.here`)"
  - "traefik.http.routers.dashboard-api.service=api@internal"

- "traefik.http.routers.dashboard-api.middlewares=dashboard-auth-user"
- "traefik.http.middlewares.dashboard-auth-user.basicauth.users=<user>:$$pws5$$rWsEfeUw9$$uV45uwsGeaPbu8RSexB9/"
- "traefik.http.routers.dashboard-api.tls.certresolver=cloudflareresolver"

- "traefik.http.middlewares.frame-deny.headers.framedeny=true"
- "traefik.http.middlewares.browser-xss-filter.headers.browserxssfilter=true"
- "traefik.http.middlewares.ssl-redirect.headers.sslredirect=true"

traefik:
  container_name: traefik
  image: "traefik:latest"
  restart: unless-stopped
  mem_limit: 40m
  mem_reservation: 25m
  ports:
    - "80:80"
    - "443:443"
  command:
    - --entrypoints.web.address=:80
    - --entrypoints.websecure.address=:443
    - --entrypoints.web.http.redirections.entryPoint.to=websecure
    - --entrypoints.web.http.redirections.entryPoint.scheme=https
    - --providers.docker
    - --log.level=INFO
    - --api.dashboard=true
    - --metrics.prometheus=true
    - --certificatesresolvers.cloudflareresolver.acme.email=<your@email.here>
    - --certificatesresolvers.cloudflareresolver.acme.dnschallenge.provider=cloudflare
    - --certificatesresolvers.cloudflareresolver.acme.storage=./acme.json
  volumes:
    - "/var/run/docker.sock:/var/run/docker.sock:ro"
    - "./traefik/acme.json:/acme.json"
  environment:
    - CLOUDFLARE_EMAIL=<your-cloudflare@email.here>
    - CLOUDFLARE_API_KEY=<your-api-key-goes-here>
  labels:
    - "traefik.http.routers.dashboard-api.rule=Host(`dashboard.your-host.here`)"
    - "traefik.http.routers.dashboard-api.service=api@internal"
    - "traefik.http.routers.dashboard-api.middlewares=dashboard-auth-user"
    - "traefik.http.middlewares.dashboard-auth-user.basicauth.users=<user>:$$pws5$$rWsEfeUw9$$uV45uwsGeaPbu8RSexB9/"
    - "traefik.http.routers.dashboard-api.tls.certresolver=cloudflareresolver"
    - "traefik.http.middlewares.frame-deny.headers.framedeny=true"
    - "traefik.http.middlewares.browser-xss-filter.headers.browserxssfilter=true"
    - "traefik.http.middlewares.ssl-redirect.headers.sslredirect=true"

nginx:
  container_name: nginx
  image: nginxinc/nginx-unprivileged:alpine
  restart: unless-stopped
  mem_limit: 8m
  command: ["nginx", "-enable-prometheus-metrics", "-g", "daemon off;"]
  volumes:
    - "./blog/:/usr/share/nginx/html/blog:ro"
    - "./nginx/default.conf.template:/etc/nginx/templates/default.conf.template:ro"
  environment:
    - NGINX_BLOG_PORT=80
    - NGINX_BLOG_HOST=<blog.your-host.here>

server {

    listen       ${NGINX_BLOG_PORT};
    server_name  localhost;

    root   /usr/share/nginx/html/${NGINX_BLOG_HOST};

    location / {
	index  index.html;
	try_files $uri $uri/ =404;
    }
}

labels:
  - "traefik.http.routers.blog-http.rule=Host(`blog.your-host.here`)"

- "traefik.http.routers.blog-http.service=blog-http"
- "traefik.http.services.blog-http.loadbalancer.server.port=80"

- "traefik.http.routers.blog-http.middlewares=blog-main"
- "traefik.http.middlewares.blog-main.chain.middlewares=frame-deny,browser-xss-filter,ssl-redirect"

- "traefik.http.routers.blog-http.tls.certresolver=cloudflareresolver"

nginx:
  container_name: nginx
  image: nginxinc/nginx-unprivileged:alpine
  restart: unless-stopped
  mem_limit: 8m
  command: ["nginx", "-enable-prometheus-metrics", "-g", "daemon off;"]
  volumes:
    - "./blog/:/usr/share/nginx/html/blog:ro"
    - "./nginx/default.conf.template:/etc/nginx/templates/default.conf.template:ro"
  environment:
    - NGINX_BLOG_PORT=80
    - NGINX_BLOG_HOST=<blog.your-host.here>
  labels:
    - "traefik.http.routers.blog-http.rule=Host(`blog.your-host.here`)"
    - "traefik.http.routers.blog-http.service=blog-http"
    - "traefik.http.services.blog-http.loadbalancer.server.port=80"
    - "traefik.http.routers.blog-http.middlewares=blog-main"
    - "traefik.http.middlewares.blog-main.chain.middlewares=frame-deny,browser-xss-filter,ssl-redirect"
    - "traefik.http.routers.blog-http.tls.certresolver=cloudflareresolver"

---
version: '2.3'

services:

  traefik:
    container_name: traefik
    image: "traefik:latest"
    restart: unless-stopped
    mem_limit: 40m
    mem_reservation: 25m
    ports:
      - "80:80"
      - "443:443"
    command:
      - --entrypoints.web.address=:80
      - --entrypoints.websecure.address=:443
      - --entrypoints.web.http.redirections.entryPoint.to=websecure
      - --entrypoints.web.http.redirections.entryPoint.scheme=https
      - --providers.docker
      - --log.level=INFO
      - --api.dashboard=true
      - --metrics.prometheus=true
      - --certificatesresolvers.cloudflareresolver.acme.email=<your@email.here>
      - --certificatesresolvers.cloudflareresolver.acme.dnschallenge.provider=cloudflare
      - --certificatesresolvers.cloudflareresolver.acme.storage=./acme.json
    volumes:
      - "/var/run/docker.sock:/var/run/docker.sock:ro"
      - "./traefik/acme.json:/acme.json"
    environment:
      - CLOUDFLARE_EMAIL=<your-cloudflare@email.here>
      - CLOUDFLARE_API_KEY=<your-api-key-goes-here>
    labels:
      - "traefik.http.routers.dashboard-api.rule=Host(`dashboard.your-host.here`)"
      - "traefik.http.routers.dashboard-api.service=api@internal"
      - "traefik.http.routers.dashboard-api.middlewares=dashboard-auth-user"
      - "traefik.http.middlewares.dashboard-auth-user.basicauth.users=<user>:$$pws5$$rWsEfeUw9$$uV45uwsGeaPbu8RSexB9/"
      - "traefik.http.routers.dashboard-api.tls.certresolver=cloudflareresolver"
      - "traefik.http.middlewares.frame-deny.headers.framedeny=true"
      - "traefik.http.middlewares.browser-xss-filter.headers.browserxssfilter=true"
      - "traefik.http.middlewares.ssl-redirect.headers.sslredirect=true"

  nginx:
    container_name: nginx
    image: nginxinc/nginx-unprivileged:alpine
    restart: unless-stopped
    mem_limit: 8m
    command: ["nginx", "-enable-prometheus-metrics", "-g", "daemon off;"]
    volumes:
      - "./blog/:/usr/share/nginx/html/blog:ro"
      - "./nginx/default.conf.template:/etc/nginx/templates/default.conf.template:ro"
    environment:
      - NGINX_BLOG_PORT=80
      - NGINX_BLOG_HOST=<blog.your-host.here>
    labels:
      - "traefik.http.routers.blog-http.rule=Host(`blog.your-host.here`)"
      - "traefik.http.routers.blog-http.service=blog-http"
      - "traefik.http.services.blog-http.loadbalancer.server.port=80"
      - "traefik.http.routers.blog-http.middlewares=blog-main"
      - "traefik.http.middlewares.blog-main.chain.middlewares=frame-deny,browser-xss-filter,ssl-redirect"
      - "traefik.http.routers.blog-http.tls.certresolver=cloudflareresolver"

docker-compose up -d

Inventory(Hosts) file contains the information of the Ansible nodes. We can group based on our use case like web servers, DB servers and uses these groups to execute commands on them. But you have to create a group in the inventory file, let see how to do that.

[webserver]
IP_ADDRESS_1
IP_ADDRESS_2
IP_ADDRESS_3

[webserverduck]
IP_ADDRESS_1

[dbserver]
IP_ADDRESS_1
IP_ADDRESS_2
IP_ADDRESS_3

[clientduck:children]
dbserver
webserverduck

You can even assign variable nodes or group wise in inventory file

[webserverduck:vars]
ansible_connection=ssh
ansible_user=myotheruser
gather_fact=no


[dbserver]
IP_ADDRESS_1 ansible_user=myuser
IP_ADDRESS_2
IP_ADDRESS_3

Cheers!

#100DaysToOffload #Ansible #dgplug

As my followers well know, by now, I am a tinkerer at heart. Why do I do things ? No one knows ! I don't even know.

All I know, all I can tell you is that I like to see what can I do with the tools I have at hand. How can I bend them to my will. Why, you may ask. The answer is a bit complicated; part of who I am, part of what I do as a DevOps. End line is, this time I was curious.

I went down a road that taught me so much more about containers, docker, docker-compose and even Linux itself.

The question I had was simple, can I run a container only through Tor running in another container?

docker run -it -p 127.0.0.1:8118:8118 -d dperson/torproxy

curl -Lx http://localhost:8118 http://jsonip.com/

---
version: '3.9'
services:
  torproxy:
    image: dperson/torproxy
    container_name: torproxy
    restart: unless-stopped

air-gapped:
  image: ubuntu
  container_name: air-gapped
  restart: unless-stopped
  command:
    - bash
    - -c
    - sleep infinity
  networks:
    - no-internet

networks:
  no-internet:
    driver: bridge
    internal: true

docker-compose up -d

docker exec air-gapped apt-get update

Err:1 http://archive.ubuntu.com/ubuntu focal InRelease
  Temporary failure resolving 'archive.ubuntu.com'
Err:2 http://security.ubuntu.com/ubuntu focal-security InRelease
  Temporary failure resolving 'security.ubuntu.com'
Err:3 http://archive.ubuntu.com/ubuntu focal-updates InRelease
  Temporary failure resolving 'archive.ubuntu.com'
Err:4 http://archive.ubuntu.com/ubuntu focal-backports InRelease
  Temporary failure resolving 'archive.ubuntu.com'
Reading package lists...
W: Failed to fetch http://archive.ubuntu.com/ubuntu/dists/focal/InRelease  Temporary failure resolving 'archive.ubuntu.com'
W: Failed to fetch http://archive.ubuntu.com/ubuntu/dists/focal-updates/InRelease  Temporary failure resolving 'archive.ubuntu.com'
W: Failed to fetch http://archive.ubuntu.com/ubuntu/dists/focal-backports/InRelease  Temporary failure resolving 'archive.ubuntu.com'
W: Failed to fetch http://security.ubuntu.com/ubuntu/dists/focal-security/InRelease  Temporary failure resolving 'security.ubuntu.com'
W: Some index files failed to download. They have been ignored, or old ones used instead.

---
version: '3.9'
services:

  torproxy:
    image: dperson/torproxy
    container_name: torproxy
    restart: unless-stopped
    networks:
      - no-internet
      - internet

  air-gapped:
    image: ubuntu
    container_name: air-gapped
    restart: unless-stopped
    command:
      - bash
      - -c
      - sleep infinity
    networks:
      - no-internet

networks:
  no-internet:
    driver: bridge
    internal: true
  internet:
    driver: bridge
    internal: false

docker-compose up -d

docker exec -it air-gapped bash

echo 'Acquire::http::Proxy "http://torproxy:8118/";' > /etc/apt/apt.conf.d/proxy
echo "export HTTP_PROXY=http://torproxy:8118/" >> ~/.bashrc
echo "export HTTPS_PROXY=http://torproxy:8118/" >> ~/.bashrc
export HTTP_PROXY=http://torproxy:8118/
export HTTPS_PROXY=http://torproxy:8118/
apt-get update
apt-get upgrade -y
DEBIAN_FRONTEND=noninteractive apt-get install -y curl

env | grep HTTP

HTTPS_PROXY=http://torproxy:8118/
HTTP_PROXY=http://torproxy:8118/

curl https://jsonip.com/

Recently, Google announced their new ad-tracking and surveillance tool called Federated Learning of Cohorts (FLoC). This is a new alternative to the third-party cookie tracking that is otherwise widely used for advertising business.

EFF has written more about the issues with using Google FLoC and also created a website where you can test if you are already a victim of their FLoC tests.

Google will track any user visiting your website even if it doesn't have Google analytics or any other services related to Google. One easy way for users visiting websites to opt out of this is to not use Google Chrome and use browsers like Firefox, etc. However, website maintainers can also help against this new tracking technology by opting out of the FLoC network.

Permissions-Policy Header

So the main way of opting out of this surveillance technology is to add a HTTP response header to their websites.

The HTTP response header is

Permissions-Policy: interest-cohort()

The FLoC technology uses interest-cohort to check for an allowlist. By default, everything is allowed as long as the user is visiting from a browser which supports InterestCohort API. However, by mentioning interest-cohort() in the Permissions-Policy header, the website is opting out from allowing any origin (including the current page) from being tracked using FLoC. Hence the FLoC feature is turned off for the website, even if the user is visiting your website from a Google Chrome browser.

NGINX

To add the above header, go to your nginx configuration file and add the following inside the server block:

server {
    ...

    add_header Permissions-Policy "interest-cohort=()";
    
    ...
}

If you have different nginx confs for multiple websites running via nginx, you have to do the above in all the server blocks or nginx configuration files.

Then run nginx -t to test that everything is correct in your updated nginx configuration. Then, restart nginx by running the command service nginx restart (or any other command that you might use based on your OS to restart nginx)

If you are using any other web server or proxy server, you can check this link: https://paramdeo.com/blog/opting-your-website-out-of-googles-floc-network

I’m trying to post these roughly once a month. Here’s the Feb post.

Work I did (1 Mar 2021 - 31 Mar 2021)

Technical

• Started writing a PEP 517 build backend for Sphinx themes.
• Started writing sphinx-basic-ng, an attempt at modernising the Sphinx theme ecosystem.
• Released new versions of sphinx-autobuild.
• More updates to Furo.
• Made progress on installer’s implementation.
• Made progress on pip’s documentation rewrite.
• Started work on rewriting ScriptTest, to pay down pip’s technical debt.

Communication

• Talking to relevant folks about toml on PyPI (and moving it to… a specific GitHub repository).
• Finally sat down and did an “open source responsibility audit”.
• My open social weekends is still a thing, and has been great so far!
• Still collaborating on designing a lockfile format for Python.
• Added to The Executable Book Project’s GitHub organisation! ^>^
• Chatted with a few people about the state of Python and pip on Debian.

General notes

Looking back, I think I picked up a couple of new projects based on random brain waves I had! That’s perfectly timed, because I’ve decided to pivot away from my earlier approach of “yay, more responsibility!”.

What next?

This is getting increasingly harder to decide on, as my free time chunks are becoming smaller and I’m picking up bigger projects. :)

Technical

• Sphinx Theme PEP 517 stuff: Make the initial release.
• sphinx-basic-ng: Make the first usable release.
• pip: Clear some of the backlog on the pull request front.
• pip: More progress on the documentation rewrite.

Communication

• Spend more time looking into the Python lockfile standardisation effort.
• Write a blog post, on automated code formatting.
• Find more speaking opportunities, to talk about things that aren’t Python packaging!

Other commitments

A full time job, that pays my bills. :)

Sure, it sounds like a click-bait headline, but I'm serious. I've bought my Android phone, a Google Nexus X5, second-hand from ebay. It's a great device, popular with developers but it's not known for its long lasting battery. In fact, I had to charge my phone at least once a day, if not more often.

But not anymore. During the last three weeks I charged it twice, no more. I didn't just turn it off. What I did is simple: I stopped using it. I turned off the WiFi, the mobile data connection and as a result I don't have much reason to look at it very often. Sometimes not even once a day. I do get phone calls from time to time, and I even play some puzzle games while I brush my teeth, but that's it.

Why no smart phone?

I would say I was an average smart phone user. I do some of that social networking thingies, message with friends and family and if I don't know the answer to a random question I ask duckduckgo. I've not been a poweruser, I don't have a zillion followers on any platform, and I don't take pictures of me or my food or my cats (don't have any) all the time. But I did use my phone regularly to check my email, my calendar or via termux work remotely on several Linux servers.

I cannot give any reliable numbers about the amount of time I've spent staring on my smart phone, but I can tell that I took it everywhere. I took it to the toilet to read my mastodon timeline and I took it with me when I went to bed to read my RSS feeds. I had it lying next to me on the table during meals or at work, so I could peek at incoming messages.

After a couple of weeks without using my smart phone now, I can tell that I miss none of that. I integrated some things into a more or less regular schedule, like checking feeds and mastodon on the laptop only once a day before I go to bed. It saves a lot of time and I never had the feeling I would miss anything. However, I find much more time to read books or play some guitar now, which I enjoy very much.

I spending a huge amount of my day in front of a screen anyway, due to my job and hobbies, but I feel that dropping the phone is absolutely beneficial for my health and also I could be the better role model for my children.

How no smart phone?

That leaves us with the question: What apps/services to drop and how to replace those that I'm not willing to drop?

The obvious

The easiest part was to port the most phone actions to the laptop. I did read mails on the phone a lot, though I didn't reply very often. Sometimes I just flagged some as important and left the work for later. Doing email on my laptop only resulted in more consistency and structure for me, and additionally saved quite some time and stress in my off-work time.

Randomly browsing the web or looking up stuff on Wikipedia, just like reading feeds and timelines was also easy to replace and led to more structure and much less distraction. Interestingly, I don't think that I'm spending more time on the laptop now, but I never took any measure to verify that ;)

I used my phone as second factor for some two-factor authentication logins. For now I also replaced that with the OTP module of my password manager pass. It still is a valid second factor, as I need the laptop additionally to the password, however, I think I will replace that with a security key in future.

I still enjoy Spotify sometimes. To make that work without my phone, I installed spotifyd on my raspberry pi, which I use as media center. The daemon just serves as the player, while I'm controlling the playlist from my laptop (using a beautiful TUI written in Rust, called spt).

The messaging

I'm using Signal as my main messaging app, but also joined some Matrix rooms on different servers and even follow some Mattermost conversations. Since I never seriously used IRC on the phone, I instead tried to integrate the messaging to my favorite IRC client weechat, with quite decent success.

I'm running signald on my weechat server and fixed some on the signal-weechat plugin and message happily ever since. Signald is a java app that subscribes to my phone number and provides a socket that can be used to interact with the API. Is it as secure as the official Signal app? Nope. But that's not my threat model ;) It's way better than the crappy electron app, in my opinion.

After I configured that, I realized that there are also plugins available for matrix and Mattermost (the latter of which is in pre-alpha, but hey). Using weechat for all my messaging feels great and to me is a big improvement over the phone (and over web- and electron clients). I know that there may be security drawbacks, not in transportation but in the way data is stored on my server. Luckily I can address these issues and I can do that independent of any app.

The rest

I thought that I could just drop everything that I cannot port to laptop. However, I hardly found anything I had to give up. I still use the phone to take pictures sometimes (never took many) and use the audio recorder occasionally. I play less random games.

The only thing I miss is the calendar functionality. I do have an excellent calendar on the laptop, or course (khal). But sometimes I would like to quickly glance at my upcoming events or schedule an appointment without being on the laptop. I will need to think how to do that without going fully analog on this one again. I'm having an eye on the Mudita Pure which I think would support me in exactly the way I chose to take.

Alrighty! Let’s start doing this again. The plan is to get back to doing these roughly once a month again.

Work I did (1 Jan 2021 - 26 Feb 2021)

Technical

• Published the last version of pip that supports Python 2! 🎉
• Published a few releases of Furo – a Sphinx theme I wrote.
• Made some (unreleased) changes to sphinx-autobuild.
• Made some more progress on installer – a reusable library for Python (wheel) package installation.
• Rewrote get-pip.py and the generation pipeline for it.
• Started a rewrite of pip’s documentation. I’d love to get some feedback on this.

Communication

• I’m experimenting with a new thing: social weekends!
• I presented 2 talks at FOSDEM: in the Python devroom and Open Source Design devroom. Shout-out to Bernard Tyers, for all the help and the bazillion reminders to make sure I do all the things on time. :)
• Collaborating on designing a lockfile format for Python, that can hopefully be standardised for interoperability.

General notes

Onboarding in a new company, relocating internationally, settling into a new space has been… well, it’s all been a very interesting learning experience.

Given the fairly strict lockdown and the percentage of people wearing masks in my locality, I’ve spent a lots of time indoors. Looking forward to the social weekends experiment I’m doing.

What next?

Technical

• pip: Work on the documentation rewrite, hopefully to get it ready in time for the next release.
• pip: Clear some of the backlog on the pull request front.
• pip: General discussions for new features and enhancements.
• TOML: Work on writing that the compliance test suite.
• TOML: Bring toml for Python back from the dead.
• Furo: Make the first stable release.
• Start work on the other Sphinx theme I have in mind.

Communication

• Spend more time looking into the Python lockfile standardisation effort.
• Catch up on the Python-on-Debian saga, and see how I can contribute constructively.

Other commitments

Oh, I have a full time job at Bloomberg now. :)

Let's dive in a little deeper into the feature.

Why do we need an anonymous chat?

A common question that we got during developing this feature is what's the use of an anonymous chat room since we already have end-to-end encrypted messaging apps. It leaves a lot fewer traces.

The way we achieve this is very simple. There is no form of storage whatsoever in OnionShare chat mode. The chat is not persistent. The chat server stores no information at all (not even the usernames of people chatting). So once the chat server is closed, and the Tor Browser tab with the chat client is closed, there is no data (or metadata) related to chat that remains, even in the person's system who started the server. Hence, it leaves much less trace compared to other chat applications.

A good example of the above as mentioned by Micah in his blog is:

If, for example, you send a message to a Signal group, a copy of your message ends up on each device (the devices, and computers if they set up Signal Desktop of each member of the group). Even if disappearing messages is turned on it’s hard to confirm all copies of the messages are actually deleted from all devices, and from any other places (like notifications databases) they may have been saved to. OnionShare chat rooms don’t store any messages anywhere, so the problem is reduced to a minimum.

Given that the OnionShare chat feature works over the onion network, so it also has the additional anonymity feature. Also, adding to the anonymity feature, OnionShare chat doesn't need any form of signing in. Hence, people chatting can stay anonymous, and everything happens inside the tor network. One can just start a chat server, share the link via some disposable way, and then wait for the other people to join while maintaining anonymity.

Because it's an onion service, there is no way for an attacker to eavesdrop on the messages. The closest they can get is if they run a malicious Tor rendezvous node that's chosen for the onion service, they'll be able to spy on encrypted onion traffic. So, there's no capturing ciphertext to decrypt later on.

So what happens under the hood?

The chat feature is dependent on flask-socketio and eventlet for the WebSocket server implementation, and socket.io client js for the frontend implementation of the chat client. So when a chat server is started, the WebSocket is started in a namespace "/chat". Whenever a new user joins the link, they are given a randomly generated username and they are added to the room "default". There is only one room, and the actual name of the room can be set from the OnionShare settings-related code, but it doesn't really impact anything in the implementation. Both the room name and the randomly generated username are stored in a flask session. But that information is also completely gone once the chat server is stopped. The room and username information are only there to emit the messages properly.

You can also change the randomly generated username to a username (or pseudo username) of your choice for that particular session.

There are two main types of messages:

1. status messages - these are sent from the client to the server only when a new user joins or someone updates their username. The status message is then broadcasted to all the other connected clients, who will then see it as a form of a status message in the chat window.

2. user messages - these are sent when a user sends a message. All messages are broadcasted, so in case you share the link to multiple users, there is no concept of private message and everyone connected to the room can view your messages. Hence, sharing the onion link securely is important.

All of these WebSocket communication happens over the Tor onion services. OnionShare in itself doesn't implement any encryption algorithm to the chat and heavily relies on the Tor onion service's encryptions for the same. The message from the client to the OnionShare server is E2EE as it goes via Tor's onion connection. Then the OnionShare server broadcasts the message to all the other clients connected to the chat room through their E2EE onion connection, over WebSockets.

So what now?

I feel, as of now, the OnionShare anonymous chat is great if you quickly want to have an anonymous, secure, non-persistent conversation with someone or a group of people. It is also great if a whistleblower wants to share some details over chat with a journalist and then remove all traces of that conversation completely. But I feel if someone needs to host a chat server for a long time where people can connect anonymously, this is probably not the best solution for that.

There are still some issues that we will hopefully improve in the next releases. Firstly, we need to try and make it a bit more asynchronous. Right now, if left inactive for a long time, the Tor connection over WebSocket is sometimes dropped which isn't a great user experience. We are working on improving that.

Also, we will improve the UI for the chat more to give a better user experience.

With the new tabs feature, one can have all different modes (even multiple servers of same mode) running at the same time. So you can have a chat server and share mode server running at the same time. All the modes are very independent of each other and hence don't affect one another in any way.

I hope you all enjoy the new chat feature and leave feedbacks/suggestions on how to improve it. You can also read more about this and other features at docs.onionshare.org

Background

I recently get really pissed quite often when I browse the web; I'm nuked with ads and pop-ups or - should I dare to use an adblocker - will not be able to see the content I was looking for at all. With the GDPR in place, living in Europe also reveals the incredible amount of Cookies used for profiling and tracking my browsing behavior. I'm being tracked everywhere, always.

Sure, I can opt-out. That's what they say. And it's a bad joke.

I cannot opt-out of what is called "essential to make the site run", which is obviously a whole lot. Technically it's forbidden to make the opt-out process harder then the opt-in, but I have never seen a single of these modal dialogs that would have offered me an "opt-out of everything" button that was just as easy to find as the "I agree to everything" button. Now even if I opt-out of everything, sites hide their "legitimate interest" settings on another page where I often need to manually de-select zillion items.

That's not the web I learned to love and I want to use as a major source of daily information.

I also recently learned about a new internet protocol that aims to make things different than HTTP/S, called gemini. If you're familiar with the gopher protocol, which has been around for some time now, you can put gemini somewhere between gopher and HTTP/S. By intention it does not much. It will let you display text, basically, that you type in a markdown-like syntax called gemtext. The function set is very limited on purpose, so content counts. It does not support images or inline links, but three levels of headings and code blocks. gemini does also TLS by default and explicitly welcomes self-signed certificates.

Hosting a Gemlog

To serve a gemlog, a server is needed; since TLS is a requirement, it's not sufficient to just serve the file structure. There are a couple of lightweight servers available, I'm using gemserv here, which I found fairly easy to compile and configure.

Generating a Certificate

I use openssl to generate a private key and a certificate signing request (CSR):

openssl req -new -newkey rsa:4096 -nodes \
  -keyout private_key.txt \
  -out csr.txt \
  -subj "/CN=gemini.schubisu.de"

This will generate an X.509 certificate signing request with a new RSA 4096 bit key. The nodes option will skip encryption of the key. You can put more information in the subject name, it's important to have the CN right, however, to let it match your domain.

With the CSR we can generate the certificate:

openssl x509 -req -sha256 \
  -days 365 \
  -in csr.txt \
  -signkey private_key.txt \
  -out certificate.txt \
  -extensions req_ext

The private_key.txt and certificate.txt will need to be on our server.

Configure gemserv

If available in your packet manager, install gemserv from there, or clone the repository and follow the build instructions. The configuration can be as simple as the following config.toml file:

port = 1965
host = "0.0.0.0"

[[server]]
hostname = "gemini.schubisu.de"
key = "/path/to/private_key.txt"
cert = "/path/to/certificate.txt"
dir = "/path/to/content"

and simply run

/path/to/gemserv config.toml

For local testing you don't even need to have a registered domain, just add a line like this to your /etc/hosts file:

127.0.0.1    gemini.schubisu.de    localhost

Converting my lektor blog to gemtext

I write my blog with a lightweight CMS for static websites written in python, called lektor. I always liked about lektor that it's file structure mirrors the resulting web page. Every page I write is represented by a folder that contains at least one content.lr file which consists of a bit yaml and markdown.

This is very much like what gemini expects, which will either show you the subfolders of a directory, or - if present - renders the index.gmi (or index.gemini) file.

I started by renaming all contents.lr files to index.gemini to get the quickest possible working gemlog and an initial idea of what work lies ahead of me. It works, of course, but there is still plenty of room for improvement. What needs to be adjusted are

• Meta data: The contents.lr starts with a yaml block of meta data which could be anything. In my case it's the title, publication date and some tags.
• Inline links and links in general: Gemini does not allow inline links, so my links need to be converted to the correct format.
• Navigation: I want to allow quick navigation to previous and next blog posts, as well as returning to an overview.

I decided to write a small python script, to use lektor's API to access all meta information and contents. I also reviewed all my blog posts to bring the links into a standardized format:

 # inline links:
 [some text][identifyer:my_link_name]

 # and at the bottom of the document:
 [identifyer:my_link_name]: https://...

The script will convert that into valid gemini links that are numerically indexed inline when rendered:

# inline links:
some text[1]

# and just below the paragraph:
[1] my_link_name [some text]

I also edited a few minor things in my contents.lr to make the markdown comply with gemtext:

• gemtext does not allow nested bullet lists, so I need to reduce lists to one level wherever I used more
• Different than with when I convert markdown to HTML, gemtext will break a line wherever I inserted a line break. If I want something to show as a nice paragraph, it should not be interrupted by line breaks.
• Inline text formatting to emphasize some words is not possible in gemtext. However, I find markdown with inline formatting still readable enough, so I left that as is.
• I found a whole lot of typos :)

I've uploaded the script to convert the blog here. It's pretty much tailored to my needs. However, it may be useful if you happen to run into the same or similar idea :)

For what it's worth, I can now simply run my conversion script, rsync the files to my gemini server and that's it.

You can grab a gemini browser and checkout the results of my conversion at gemini://gemini.schubisu.de

During this period last year, I got interested in how a new Kubernetes version is released and what goes on behind it. After some searching, I found that all of the process and the roles are well documented in the Release Team Role Handbooks.

I read through all of them to understand the process, why there are so many roles and what responsibilities each team is entrusted with. All of that sounded pretty interesting to me. All the teams did an amazing work and were equally crucial for the well functioning of a Kubernetes Release cycle. I got specifically interested with the Enhancements and CI Signal teams. I started to dig how can I lend my hand to the efforts.

Shadow Roles

With the role handbooks, I got to know of the Release Team Shadow Program which is aimed at mentoring new contributors and training them to be the next leads of the Kubernetes Release Team. The shadows are expected to learn from the leads and fill in wherever necessary. You can think of these positions as “trainee/intern” roles at your workplace. This was just a primer on the program. You can read more on the link.

Okay. I know where to start with. But how?

Turns out it requires a marginal amount of effort, bucket loads of curiosity and time commitment to apply for the Shadow Program. The Release Team at the start of every release cycle pushes out a public form for inviting applications to the shadow roles.

The #Kubernetes 1.17 Release Team Shadow Application is open!

If you're interested in joining one of the most challenging, but rewarding parts of the community, please apply.

No experience is necessary, outside of a desire to learn! :)

Closes 9/27.https://t.co/zm9r3MxCPw

— Stephen Augustus (he/him) (@stephenaugustus) September 19, 2019

Taking the plunge

I took the initiative and filled the form with my interests and thoughts. And few days later I, along with other shadows, was welcomed by MrBobbyTables to my first involvement with the Release Team. 🎉

The next few months were just like a roller coaster ride. The team that I was shadowing for was the Enhancements Team and our work was to shepherd features for the Kubernetes release and maintain the status of Kubernetes Enhancements Proposals, aka, KEP(s).

The role involved understanding each outstanding KEP, pinging respective OWNERS if the enhancement would be graduating in the current release cycle, and keeping track whether the enhancements are satisfying the requirements for the release.

Here were a few takeaways that I took while working on the team:

• Knowledge of what is involved when adding a feature into Kubernetes!!!
• Reading through an enourmous amount of KEPs, I got to know about the features themselves
• Communicating effectively with others and breaking the ice
• A lot of GitHub triage skills and tricks
• Tricks of wrangling data on a spreadsheet 😉

I will write about the complete lifecycle of a Kubernetes Enhancement Proposal in a future article.

We released Kubernetes 1.17 : The Chillest Release (Yes! That is the release theme 😃) after all the efforts of the 1.17 release team. The last release of the year is usually the most chilled out and is a bit short due to the December vacation.

Excited to announce the delivery of #Kubernetes 1.17, our fourth and final release of 2019! 🌟🎉🥳 https://t.co/Wf6Ku8eGZg

— Kubernetes (@kubernetesio) December 10, 2019

The whole team became akin to a family for me spending all that effort into ensuring a smooth release. We, the Enhancements Team (Bob, Jeremy, Anna, Kristin), even got together at KubeCon San Diego to meet up physically.

The Next Steps

After Kubernetes 1.17, I signed up again for the Kubernetes 1.18 team for Enhancements to get more exposure to the KEP landscape. This time Jeremy was leading the Enhancements Team. It was fun again to work with the enhancements team (Jeremy, Kirsten, Heba, John), all the same just this time there were new shadows along with me compared to last time.

This release cycle was mostly the same for me other than I was a bit more involved than the last time having served on the team previously. And, after all the hard work of the 1.18 release team, we were treated to a quarky Kubernetes 1.18 ⭐.

And Kubernetes 1.18.0 is live! 👏

Cheers to all the Release Team members who made it possible.

Read the announcement to get highlights of the 1.18 release.https://t.co/FwKpQ4nfPq@kubernetesio @CloudNativeFdn

PS. Be "A bit quarky" 🎇

(Logo by @marulango) pic.twitter.com/FsgejYAa0j

— Nabarun Pal (@theonlynabarun) March 25, 2020

Graduating to be the Enhancements Lead 🚀

Aaaaaand after a splendid 1.18, Jeremy nominated me to be the Enhancements Lead for Kubernetes 1.19 Release Team. I was stoked to get the opportunity and at the same time scared if I can do full justice to the responsibility bestowed upon me. The knowledge of the role while working with Bob and Jeremy in the previous release teams gave me the confidence that I can fulfill the responsibilities of the role.

This release cycle eventually became special in many ways. We were hit by a deadly pandemic which changed a lot of things in our life. The release cycle was extended to 5 months instead of the usual 12 weeks cadence. The pandemic and various other factors shaved off quite a bit of the usual bandwidth the community had previously. These times were very crucial for the whole world and the team didn’t want to put undue additional pressure on the amazing contributors that we have.

The shadows that I selected for the Enhancements Team spanned 12 and a half hours of timezone and created an amazing round the earth coverage for the team. This meant no team member had to toil in their odd times of the day. I take this opportunity to thank again all the enhancements shadows - Kirsten, Harsha, Miroslaw and John for their efforts even in such hard times.

After those tense & tough 5 months firefighting a lot of issues, the 1.19 release team released Kubernetes 1.19: Accentuate the Paw-sitive. You can read about the release on the Kubernetes Blog and the upcoming Kubernetes 1.19 Release Webinar where I will be presenting along with the 1.19 Release Lead Taylor and the 1.19 Communications Lead Max.

Wow! @kubernetesio v1.19.0 is out!!! This release is named, “Accentuate the Paw-sitive” for the attitude everyone had throughout the release process. A thank you to @emanate_design for designing the release logo, and a HUGE thank you to everyone who made this release possible. pic.twitter.com/20onqgG29X

— Taylor Dolezal (@onlydole) August 26, 2020

With that, my watch ended over the Release Enhancements Team and it was time to hand over the baton to next lead. ☮

I was very happy to nominate Kirsten to succeed me as the next Enhancements Lead of the Release Team. 🎉

Along with that it was time for me to graduate to my next role. I look forward to working with Jeremy, Savitha and Daniel for Kubernetes 1.20. I will be shadowing the Release Lead for Kubernetes 1.20. 😎

How can you get involved? 🙌

The last release of the year, Kubernetes 1.20, is going to be published in December. The Release Team is looking out for folks for the shadow roles. All you need to do is go ahead, read the role handbooks, figure out which role interests you the best and then fill the form. We ask every prospect to fill the form because we want to know if the release team would be a good fit for you and to find the right role for you in the Release Team.

The applications will be open until End of Day Friday, September 11, 2020 Pacific Time.

👀👉Interested in serving on the #Kubernetes 1.20 release team? Good news is that shadow selection is now open! Check it out and apply here - https://t.co/Fa9RkHJZBp

— Lachlan Evenson (@LachlanEvenson) August 31, 2020

Still in doubt? 🔍

I would say just go ahead and volunteer for the shadow roles. 🚢

Feel free to contact me on Twitter at @theonlynabarun, or on the Kubernetes Slack, in case you have anything to ask.

Quick References

• The selection process
• Shadow overview
• Role Handbooks

Update since the original version

I wrote this article over 6 months back and since then led the Kubernetes 1.21 Release Team. I plan to write about my experience leading the Release Team in the near future. Do subscribe to the RSS feed for updates.

I recently read this paper titled, Understanding Real-World Concurrency Bugs in Go (PDF), that studies concurrency bugs in Golang and comments on the new primitives for messages passing that the language is often known for.

I am not a very good Go programmer, so this was an informative lesson in various ways to achieve concurrency and synchronization between different threads of execution. It is also a good read for experienced Go developers as it points out some important gotchas to look out for when writing Go code. The fact that it uses real world examples from well known projects like Docker, Kubernetes, gRPC-Go, CockroachDB, BoltDB etc. makes it even more fun to read!

The authors analyzed a total of 171 concurrency bugs from several prominent Go open source projects and categorized them in two orthogonal dimensions, one each for the cause of the bug and the behavior. The cause is split between two major schools of concurrency

Along the cause dimension, we categorize bugs into those that are caused by misuse of shared memory and those caused by misuse of message passing

and the behavior dimension is similarly split into

we separate bugs into those that involve (any number of ) goroutines that cannot proceed (we call themblocking bugs) and those that do not involve any blocking (non-blocking bugs)

Interestingly, they chose the behavior to be blocking instead of deadlock since the former implies that atleast one thread of execution is blocked due to some concurrency bug, but the rest of them might continue execution, so it is not a deadlock situation.

Go has primitive shared memory protection mechanisms like Mutex, RWMutex etc. with a caveat

Write lock requests in Go have ahigher privilege than read lock requests.

as compared to pthread in C. Go also has a new primitive called sync.Once that can be used to guarantee that a function is executed only once. This can be useful in situations where some callable is shared across multiple threads of execution but it shouldn't be called more than once. Go also has sync.WaitGroups , which is similar to pthread_join to wait for various threads of executioun to finish executing.

Go also uses channels for the message passing between different threads of executions called Goroutunes. Channels can be buffered on un-buffered (default), the difference between them being that in a buffered channel the sender and receiver don't block on each other (until the buffered channel is full).

The study of the usage patterns of these concurrency primitives in various code bases along with the occurence of bugs in the codebase concluded that even though message passing was used at fewer places, it accounted for a larger number of bugs(58%).

Implication 1:With heavier usages of goroutines and newtypes of concurrency primitives, Go programs may potentiallyintroduce more concurrency bugs

Also, interesting to note is this observation in tha paper

Observation 5:All blocking bugs caused by message passing are related to Go’s new message passing semantics like channel. They can be difficult to detect especially when message passing operations are used together with other synchronization mechanisms

The authors also talk about various ways in which Go runtime can detect some of these concurrency bugs. Go runtime includes a deadlock detector which can detect when there are no goroutunes running in a thread, although, it cannot detect all the blocking bugs that authors found by manual inspection.

For shared memory bugs, Go also includes a data race detector which can be enbaled by adding -race option when building the program. It can find races in memory/data shared between multiple threads of execution and uses happened-before algorithm underneath to track objects and their lifecycle. Although, it can only detect a part of the bugs discovered by the authors, the patterns and classification in the paper can be leveraged to improve the detection and build more sophisticated checkers.

TLDR; Trying to learn new things I tried writing a URL shortner called shorty. This is a first draft and I am trying to approach it from first principle basis. Trying to break down everything to the simplest component.

I decided to write my own URL shortner and the reason for doing that was to dive a little more into golang and to learn more about systems. I have planned to not only document my learning but also find and point our different ways in which this application can be made scalable, resilient and robust.

A high level idea is to write a server which takes the big url and return me a short url for the same. I have one more requirement where I do want to provide a slug i.e a custom short url path for the same. So for some links like https://play.google.com/store/apps/details?id=me.farhaan.bubblefeed, I want to have a url like url.farhaan.me/linktray which is easy to remember and distribute.

The way I am thinking to implement this is by having two components, I want a CLI interface which talks to my Server. I don’t want a fancy UI for now because I want it to be exclusively be used through terminal. A Client-Server architecture, where my CLI client sends a request to the server with a URL and an optional slug. If a slug is present URL will have that slug in it and if it doesn’t it generates a random string and make the URL small. If you see from a higher level it’s not just a URL shortner but also a URL tagger.

The way a simple url shortner works:

Flow Diagram

A client makes a request to make a given URL short, server takes the URL and stores it to the database, server then generates a random string and maps the URL to the string and returns a URL like url.farhaan.me/<randomstring>.

Now when a client requests to url.farhaan.me/<randomstring>, it goest to the same server, it searches the original URL and redirects the request to a different website.

The slug implementation part is very straightforward, where given a word, I might have to search the database and if it is already present we raise an error but if it isn’t we add it in the database and return back the URL.

One optimization, since it’s just me who is going to use this, I can optimize my database to see if the long URL already exists and if it does then no need to create a new entry. But this should only happen in case of random string and not in case of slugs. Also this is a trade off between reducing the redundancy and latency of a request.

But when it comes to generating a random string, things get a tiny bit complicated. This generation of random strings, decides how many URLs you can store. There are various hashing algorithms that I can use to generate a string I can use md5, base10 or base64. I also need to make sure that it gives a unique hash and not repeated ones.

Unique hash can be maintained using a counter, the count either can be supplied from a different service which can help us to scale the system better or it can be internally generated, I have used database record number for the same.

If you look at this on a system design front. We are using the same Server to take the request and generate the URL and to redirect the request. This can be separated into two services where one service is required to generate the URL and the other just to redirect the URL. This way we increase the availability of the system. If one of the service goes down the other will still function.

The next step is to write and integrate a CLI system to talk to the server and fetch the URL. A client that can be used for an end user. I am also planning to integrate a caching mechanism in this but not something out of the shelf rather write a simple caching system with some cache eviction policy and use it.

Till then I will be waiting for the feedback. Happy Hacking.

I now have a Patreon open so that you folks can support me to do this stuff for longer time and sustain myself too. So feel free to subscribe to me and help me keeping doing this with added benefits.

TLDR; Link Tray is a utility we recently wrote to curate links from different places and share it with your friends. The blogpost has technical details and probably some productivity tips.

Link Bubble got my total attention when I got to know about it, I felt it’s a very novel idea, it helps to save time and helps you to curate the websites you visited. So on the whole, and believe me I am downplaying it when I say Link Bubble does two things:

1. Saves time by pre-opening the pages
2. Helps you to keep a track of pages you want to visit

It’s a better tab management system, what I felt weird was building a whole browser to do that. Obviously, I am being extremely naive when I am saying it because I don’t know what it takes to build a utility like that.

Now, since they discontinued it for a while and I never got a chance to use it. I thought let me try building something very similar, but my use case was totally different. Generally when I go through blogs or articles, I open the links mentioned in a different tab to come back to them later. This has back bitten me a lot of time because I just get lost in so many links.

I thought if there is a utility which could just capture the links on the fly and then I could quickly go through them looking at the title, it might ease out my job. I bounced off the same idea across to Abhishek and we ended up prototyping LinkTray.

Our first design was highly inspired by facebook messenger but instead of chatheads we have links opened. If you think about it the idea feels very beautiful but the design is “highly” not scalable. For example if you have as many as 10 links opened we had trouble in finding our links of interest which was a beautiful design problems we faced.

We quickly went to the whiteboard and put up a list of requirements, first principles; The ask was simple:

1. To share multiple links with multiple people with least transitions
2. To be able to see what you are sharing
3. To be able to curate links (add/remove/open links)

We took inspiration from an actual Drawer where we flick out a bunch of links and go through them. In a serendipitous moment the design came to us and that’s how link tray looks like the way it looks now.

Link Tray

Link Tray was a technical challenge as well. There is a plethora of things I learnt about the Android ecosystem and application development that I knew existed but never ventured into exploring it.

Link Tray is written in Java, and I was using a very loosely maintained library to get the overlay activity to work. Yes, the floating activity or application that we see is called an overlay activity, this allows the application to be opened over an already running application.

The library that I was using doesn’t have support for Android O and above. To figure that out it took me a few nights , also because I was hacking on the project during nights . After reading a lot of GitHub issues I figured out the problem and put in the support for the required operating system.

One of the really exciting features that I explored about Android is Services. I think I might have read most of the blogs out there and all the documentation available and I know that I still don't know enough. I was able to pick enough pointers to make my utility to work.

Just like Uncle Bob says make it work and then make it better. There was a persistent problem, the service needs to keep running in the background for it to work. This was not a functional issue but it was a performance issue for sure and our user of version 1.0 did have a problem with it. People got mislead because there was constant notification that LinkTray is running and it was annoying. This looked like a simple problem on the face but was a monster in the depth.

Architecture of Link Tray

The solution to the problem was simple stop the service when the tray is closed, and start the service when the link is shared back to link tray. Tried, the service did stop but when a new link was shared the application kept crashing. Later I figured out the bound service that is started by the library I am using is setting a bound flag to True but when they are trying to reset this flag , they were doing at the wrong place, this prompted me to write this StackOverflow answer to help people understand the lifecycle of service. Finally after a lot of logs and debugging session I found the issue and fixed it. It was one of the most exciting moment and it help me learn a lot of key concepts.

The other key learning, I got while developing Link Tray was about multi threading, what we are doing here is when a link is shared to link tray, we need the title of the page if it has and favicon for the website. Initially I was doing this on the main UI thread which is not only an anti-pattern but also a usability hazard. It was a network call which blocks the application till it was completed, I learnt how to make a network call on a different thread, and keep the application smooth.

Initially approach was to get a webview to work and we were literally opening the links in a browser and getting the title and favicon out, this was a very heavy process. Because we were literally spawning a browser to get information about links, in the initial design it made sense because we were giving an option to consume the links. Over time our design improved and we came to a point where we don’t give the option to consume but to curate. Hence we opted for web scraping, I used custom headers so that we don’t get caught by robot.txt. And after so much of effort it got to a place where it is stable and it is performing great.

It did take quite some time to reach a point where it is right now, it is full functional and stable. Do give it a go if you haven’t, you can shoot any queries to me.

Link to Link Tray: https://play.google.com/store/apps/details?id=me.farhaan.bubblefeed

Happy Hacking!

http://bikeshed.org/

What color should I paint the bike-shed?

[Published in Open Source For You (OSFY) magazine, October 2017 edition.]

This article is the eighth in the DevOps series. In this issue, we shall learn to set up Docker in the host system and use it with Ansible.

Introduction

Docker provides operating system level virtualisation in the form of containers. These containers allow you to run standalone applications in an isolated environment. The three important features of Docker containers are isolation, portability and repeatability. All along we have used Parabola GNU/Linux-libre as the host system, and executed Ansible scripts on target Virtual Machines (VM) such as CentOS and Ubuntu.

Docker containers are extremely lightweight and fast to launch. You can also specify the amount of resources that you need such as CPU, memory and network. The Docker technology was launched in 2013, and released under the Apache 2.0 license. It is implemented using the Go programming language. A number of frameworks have been built on top of Docker for managing these cluster of servers. The Apache Mesos project, Google’s Kubernetes, and the Docker Swarm project are popular examples. These are ideal for running stateless applications and help you to easily scale them horizontally.

Setup

The Ansible version used on the host system (Parabola GNU/Linux-libre x86_64) is 2.3.0.0. Internet access should be available on the host system. The ansible/ folder contains the following file:

ansible/playbooks/configuration/docker.yml

Installation

The following playbook is used to install Docker on the host system:

---
- name: Setup Docker
  hosts: localhost
  gather_facts: true
  become: true
  tags: [setup]

  tasks:
    - name: Update the software package repository
      pacman:
        update_cache: yes

    - name: Install dependencies
      package:
        name: "{{ item }}"
        state: latest
      with_items:
        - python2-docker
        - docker

    - service:
        name: docker
        state: started

    - name: Run the hello-world container
      docker_container:
        name: hello-world
        image: library/hello-world

The Parabola package repository is updated before proceeding to install the dependencies. The python2-docker package is required for use with Ansible. Hence, it is installed along with the docker package. The Docker daemon service is then started and the library/hello-world container is fetched and executed. A sample invocation and execution of the above playbook is shown below:

$ ansible-playbook playbooks/configuration/docker.yml -K --tags=setup
SUDO password: 

PLAY [Setup Docker] *************************************************************

TASK [Gathering Facts] **********************************************************
ok: [localhost]

TASK [Update the software package repository] ***********************************
changed: [localhost]

TASK [Install dependencies] *****************************************************
ok: [localhost] => (item=python2-docker)
ok: [localhost] => (item=docker)

TASK [service] ******************************************************************
ok: [localhost]

TASK [Run the hello-world container] ********************************************
changed: [localhost]

PLAY RECAP **********************************************************************
localhost                  : ok=5    changed=2    unreachable=0    failed=0   

With verbose ’-v’ option to ansible-playbook, you will see an entry for LogPath, such as /var/lib/docker/containers//-json.log. In this log file you will see the output of the execution of the hello-world container. This output is the same when you run the container manually as shown below:

$ sudo docker run hello-world

Hello from Docker!

This message shows that your installation appears to be working correctly.

To generate this message, Docker took the following steps:
 1. The Docker client contacted the Docker daemon.
 2. The Docker daemon pulled the "hello-world" image from the Docker Hub.
 3. The Docker daemon created a new container from that image which runs the
    executable that produces the output you are currently reading.
 4. The Docker daemon streamed that output to the Docker client, which sent it
    to your terminal.

To try something more ambitious, you can run an Ubuntu container with:
 $ docker run -it ubuntu bash

Share images, automate workflows, and more with a free Docker ID:
 https://cloud.docker.com/

For more examples and ideas, visit:
 https://docs.docker.com/engine/userguide/

Example

A Deep Learning (DL) Docker project is available (https://github.com/floydhub/dl-docker) with support for frameworks, libraries and software tools. We can use Ansible to build the entire DL container from the source code of the tools. The base OS of the container is Ubuntu 14.04, and will include the following software packages:

• Tensorflow
• Caffe
• Theano
• Keras
• Lasagne
• Torch
• iPython/Jupyter Notebook
• Numpy
• SciPy
• Pandas
• Scikit Learn
• Matplotlib
• OpenCV

The playbook to build the DL Docker image is given below:

- name: Build the dl-docker image
  hosts: localhost
  gather_facts: true
  become: true
  tags: [deep-learning]

  vars:
    DL_BUILD_DIR: "/tmp/dl-docker"
    DL_DOCKER_NAME: "floydhub/dl-docker"

  tasks:
    - name: Download dl-docker
      git:
        repo: https://github.com/saiprashanths/dl-docker.git 
        dest: "{{ DL_BUILD_DIR }}"

    - name: Build image and with buildargs
      docker_image:
         path: "{{ DL_BUILD_DIR }}"
         name: "{{ DL_DOCKER_NAME }}"
         dockerfile: Dockerfile.cpu
         buildargs:
           tag: "{{ DL_DOCKER_NAME }}:cpu"

We first clone the Deep Learning docker project sources. The docker_image module in Ansible helps us to build, load and pull images. We then use the Dockerfile.cpu file to build a Docker image targeting the CPU. If you have a GPU in your system, you can use the Dockerfile.gpu file. The above playbook can be invoked using the following command:

$ ansible-playbook playbooks/configuration/docker.yml -K --tags=deep-learning

Depending on the CPU and RAM you have, it will take considerable amount of time to build the image with all the software. So be patient!

Jupyter Notebook

The built dl-docker image contains Jupyter notebook which can be launched when you start the container. An Ansible playbook for the same is provided below:

- name: Start Jupyter notebook
  hosts: localhost
  gather_facts: true
  become: true
  tags: [notebook]

  vars:
    DL_DOCKER_NAME: "floydhub/dl-docker"

  tasks:
    - name: Run container for Jupyter notebook
      docker_container:
        name: "dl-docker-notebook"
        image: "{{ DL_DOCKER_NAME }}:cpu"
        state: started
        command: sh run_jupyter.sh

You can invoke the playbook using the following command:

$ ansible-playbook playbooks/configuration/docker.yml -K --tags=notebook

The Dockerfile already exposes the port 8888, and hence you do not need to specify the same in the above docker_container configuration. After you run the playbook, using the ‘docker ps’ command on the host system, you can obtain the container ID as indicated below:

$ sudo docker ps
CONTAINER ID        IMAGE                    COMMAND               CREATED             STATUS              PORTS                NAMES
a876ad5af751        floydhub/dl-docker:cpu   "sh run_jupyter.sh"   11 minutes ago      Up 4 minutes        6006/tcp, 8888/tcp   dl-docker-notebook

You can now login to the running container using the following command:

$ sudo docker exec -it a876 /bin/bash

You can then run an ‘ifconfig’ command to find the local IP address (“172.17.0.2” in this case), and then open http://172.17.0.2:8888 in a browser on your host system to see the Jupyter Notebook. A screenshot is shown in Figure 1:

TensorBoard

TensorBoard consists of a suite of visualization tools to understand the TensorFlow programs. It is installed and available inside the Docker container. After you login to the Docker container, at the root prompt, you can start Tensorboard by passing it a log directory as shown below:

# tensorboard --logdir=./log

You can then open http://172.17.0.2:6006/ in a browser on your host system to see the Tensorboard dashboard as shown in Figure 2:

Docker Image Facts

The docker_image_facts Ansible module provides useful information about a Docker image. We can use it to obtain the image facts for our dl-docker container as shown below:

- name: Get Docker image facts
  hosts: localhost
  gather_facts: true
  become: true
  tags: [facts]

  vars:
    DL_DOCKER_NAME: "floydhub/dl-docker"

  tasks:
    - name: Get image facts
      docker_image_facts:
        name: "{{ DL_DOCKER_NAME }}:cpu"

The above playbook can be invoked as follows:

$ ANSIBLE_STDOUT_CALLBACK=json ansible-playbook playbooks/configuration/docker.yml -K --tags=facts 

The ANSIBLE_STDOUT_CALLBACK environment variable is set to ‘json’ to produce a JSON output for readability. Some important image facts from the invocation of the above playbook are shown below:

"Architecture": "amd64", 
"Author": "Sai Soundararaj <saip@outlook.com>", 

"Config": {

"Cmd": [
   "/bin/bash"
], 

"Env": [
   "PATH=/root/torch/install/bin:/root/caffe/build/tools:/root/caffe/python:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin", 
   "CAFFE_ROOT=/root/caffe", 
   "PYCAFFE_ROOT=/root/caffe/python", 
   "PYTHONPATH=/root/caffe/python:", 
   "LUA_PATH=/root/.luarocks/share/lua/5.1/?.lua;/root/.luarocks/share/lua/5.1/?/init.lua;/root/torch/install/share/lua/5.1/?.lua;/root/torch/install/share/lua/5.1/?/init.lua;./?.lua;/root/torch/install/share/luajit-2.1.0-beta1/?.lua;/usr/local/share/lua/5.1/?.lua;/usr/local/share/lua/5.1/?/init.lua", 
   "LUA_CPATH=/root/torch/install/lib/?.so;/root/.luarocks/lib/lua/5.1/?.so;/root/torch/install/lib/lua/5.1/?.so;./?.so;/usr/local/lib/lua/5.1/?.so;/usr/local/lib/lua/5.1/loadall.so", 
   "LD_LIBRARY_PATH=/root/torch/install/lib:", 
   "DYLD_LIBRARY_PATH=/root/torch/install/lib:"
], 

"ExposedPorts": {
   "6006/tcp": {}, 
   "8888/tcp": {}
}, 

"Created": "2016-06-13T18:13:17.247218209Z", 
"DockerVersion": "1.11.1", 

"Os": "linux", 

"task": { "name": "Get image facts" }

You are encouraged to read the ‘Getting Started with Docker’ user guide available at http://docs.ansible.com/ansible/latest/guide_docker.html to know more about using Docker with Ansible.

[NOTE: This post originally appeared on deepsource.io, and has been posted here with due permission.]

In the early part of the last century, when David Hilbert was working on stricter formalization of geometry than Euclid, Georg Cantor had worked out a theory of different types of infinities, the theory of sets. This theory would soon unveil a series of confusing paradoxes, leading to a crisis in the Mathematics community  regarding the stability of the foundational principles of the math of that time.

Central to these paradoxes was the Russell’s paradox (or more generally, as we’d talk about later, the Epimenides Paradox). Let’s see what it is.

In those simpler times, you were allowed to define a set if you could describe it in English. And, owing to mathematicians’ predilection for self-reference, sets could contain other sets.

Russell then, came up with this:

\(R\)  is a set of all the sets which do not contain themselves.

The question was "Does \(R \) contain itself?" If it doesn’t, then according to the second half of the definition it should. But if it does, then it no longer meets the definition.

The same can symbolically be represented as:

Let \(R = \{ x \mid x \not \in x \} \), then \(R \in R \iff R \not \in R \)

Cue mind exploding.

“Grelling’s paradox” is a startling variant which uses adjectives instead of sets. If adjectives are divided into two classes, autological (self-descriptive) and heterological (non-self-descriptive), then, is ‘heterological’ heterological? Try it!

Epimenides Paradox

Or, the so-called Liar Paradox was another such paradox which shred apart whatever concept of ‘computability’ was, at that time - the notion that things could either be true or false.

Epimenides was a Cretan, who made one immortal statement:

“All Cretans are liars.”

If all Cretans are liars, and Epimenides was a Cretan, then he was lying when he said that “All Cretans are liars”. But wait, if he was lying then, how can we ‘prove’ that he wasn’t lying about lying? Ein?

⊕

This is what makes it a paradox: A statement so rudely violating the assumed dichotomy of statements into true and false, because if you tentatively think it’s true, it backfires on you and make you think that it is false. And a similar backfire occurs if you assume that the statement is false. Go ahead, try it!

If you look closely, there is one common culprit in all of these paradoxes, namely ‘self-reference’. Let’s look at it more closely.

Strange Loopiness

If self-reference, or what Douglas Hofstadter - whose prolific work on the subject matter has inspired this blog post - calls ‘Strange Loopiness’ was the source of all these paradoxes, it made perfect sense to just banish self-reference, or anything which allowed it to occur. Russell and Whitehead, two rebel mathematicians of the time, who subscribed to this point of view, set forward and undertook the mammoth exercise, namely “Principia Mathematica”, which we as we will see in a little while, was utterly demolished by Gödel’s findings.

The main thing which made it difficult to ban self-reference was that it was hard to pin point where exactly did the self-reference occur. It may as well be spread out over several steps, as in this ‘expanded’ version of Epimenides:

The next statement is a lie.

The previous statement is true.

Russell and Whitehead, in P.M. then, came up with a multi-hierarchy set theory to deal with this. The basic idea was that a set of the lowest ‘type’ could only contain ‘objects’ as members (not sets). A set of the next type could then only either contain objects, or sets of lower types. This, implicitly banished self-reference.

Since, all sets must have a type, a set ‘which contains all sets which are not members of themselves’ is not a set at all, and thus you can say that Russell’s paradox was dealt with.

Similarly, if an attempt is made towards applying the expanded Epimenides to this theory, it must fail as well, for the first sentence to make a reference to the second one, it has to be hierarchically above it - in which case, the second one can’t loop back to the first one.

Thirty one years after David Hilbert set before the academia to rigorously demonstrate that the system defined in Principia Mathematica was both consistent (contradiction-free) and complete (i.e. every true statement could be evaluated to true within the methods provided by P.M.), Gödel published his famous Incompleteness Theorem. By importing the Epimenides Paradox right into the heart of P.M., he proved that not just the axiomatic system developed by Russell and Whitehead, but none of the axiomatic systems whatsoever were complete without being inconsistent.

Clear enough, P.M. lost it’s charm in the realm of academics.

Before Gödel’s work too, P.M. wasn’t particularly loved as well.

Why?

It isn’t just limited to this blog post, but we humans, in general, have a diet for self-reference - and this quirky theory severely limits our ability to abstract away details - something which we love, not only as programmers, but as linguists too - so much so, that the preceding paragraph, “It isn’t … this blog … we humans …” would be doubly forbidden because the ‘right’ to mention ‘this blog post’ is limited only to something which is hierarchically above blog posts, ‘metablog-posts’. Secondly, me (presumably a human) belonging to the class ‘we’ can’t mention ‘we’ either.

Since, we humans, love self-reference so much, let’s discuss some ways in which it can be expressed in written form.

One way of making such a strange loop, and perhaps the ‘simplest’ is using the word ‘this’. Here:

• This sentence is made up of eight words.
• This sentence refers to itself, and is therefore useless.
• This blog post is so good.
• This sentence conveys you the meaning of ‘this’.
• This sentence is a lie. (Epimenides Paradox)

Another amusing trick for creating a self-reference without using the word ‘this sentence’ is to quote the sentence inside itself.

Someone may come up with:

The sentence ‘The sentence contains five words’ contains five words.

But, such an attempt must fail, for to quote a finite sentence inside itself would mean that the sentence is smaller than itself. However, infinite sentences can be self-referenced this way.

The sentence
    "The sentence
        "The sentence
                                ...etc
                                ...etc
        is infinitely long"
    is infinitely long"
is infinitely long"

There’s a third method as well, which you already saw in the title - the Quine method. The term ‘Quine’ was coined by Douglas Hofstadter in his book “Gödel Escher, Bach” (which heavily inspires this blog post). When using this, the self-reference is ‘generated’ by describing a typographical entity, isomorphic to the quine sentence itself. This description is carried in two parts - one is a set of ‘instructions’ about how to ‘build’ the sentence, and the other, the ‘template’ contains information about the construction materials required.

The Quine version of Epimenides would be:

“yields falsehood when preceded by it’s quotation” yields falsehood when preceded by it’s quotation

Before going on with ‘quining’, let’s take a moment and realize how awfully powerful our cognitive capacities are, and what goes in our head when a cognitive payload full of self-references is delivered - in order to decipher it, we not only need to know the language, but also need to work out the referent of the phrase analogous to ‘this sentence’ in that language. This parsing depends on our complex, yet totally assimilated ability to handle the language.

The idea of referring to itself is quite mind-blowing, and we keep doing it all the time — perhaps, why it feels so ‘easy’ for us to do so. But, we aren’t born that way, we grow that way. This could better be realized by telling someone much younger “This sentence is wrong.”. They’d probably be confused - What sentence is wrong?. The reason why it’s so simple for self-reference to occur, and hence allow paradoxes, in our language, is well, our language. It allows our brain to do the heavy lifting of what the author is trying to get through us, without being verbose.

Back to Quines.

Reproducing itself

Now, that we are aware of how ‘quines’ can manifest as self-reference, it would be interesting to see how the same technique can be used by a computer program to ‘reproduce’ itself.

To make it further interesting, we shall choose the language most apt for the purpose - brainfuck:

>>>>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++>++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++>++++++++[->++++++++<]>--....[-]<<[<]<<++++++++[->+++++>++++++++<<]>+++>-->[[-<<<+>.>>]<.[->+<]<[->+<]>>>]<<<<[<]>[.>]

Running that program above produces itself as the output. I agree, it isn’t the most descriptive program in the world, so written in Python below, is the nearest we can go to describe what’s happening inside those horrible chains of +’s and >’s:

THREE_QUOTES = '"' * 3

def eniuq(template): print(
  f'{template}({THREE_QUOTES}{template}{THREE_QUOTES})')

eniuq("""THREE_QUOTES = '"' * 3

def eniuq(template): print(
  f'{template}({THREE_QUOTES}{template}{THREE_QUOTES})')

eniuq""")

The first line generates """ on the fly, which marks multiline strings in Python.

Next two lines define the eniuq function, which prints the argument template twice - once, plain and then surrounded with triple quotes.

The last 4 lines cleverly call this function so that the output of the program is the source code itself.

Since we are printing in an order opposite of quining, the name of the function is ‘quine’ reversed -> eniuq (name stolen from Hofstadter again)

Remember the discussion about how self-reference capitalizes on the processor? What if ‘quining’ was a built-in feature of the language, providing what we in programmer lingo call ‘syntactic sugar’?

Let’s assume that an asterisk, * in the brainfuck interpreter would copy the instructions before executing them, what would then be the output of the following program?

*

It’d be an asterisk again. You could make an argument that this is silly, and should be counted as ‘cheating’. But, it’s the same as relying on the processor, like using “this sentence” to refer to this sentence - you rely on your brain to do the inference for you.

What if eniuq was a builtin keyword in Python? A perfect self-rep was then just be a call away:

eniuq('eniuq')

What if quine was a verb in the English language? We could reduce a lot of explicit cognitive processes required for inference. The Epimenides paradox would then be:

“yields falsehood if quined” yields falsehood if quined

Now, that we are talking about self-rep, here’s one last piece of entertainment for you.

The Tupper’s self-referential formula

This formula is defined through an inequality:

\({1 \over 2} < \left\lfloor \mathrm{mod}\left(\left\lfloor {y \over 17} \right\rfloor 2^{-17 \lfloor x \rfloor - \mathrm{mod}(\lfloor y\rfloor, 17)},2\right)\right\rfloor\)

If you take that absurd thing above, and move around in the cartesian plane for the coordinates \(0 \le x \le 106, k \le y \le k + 17\), where \(k\) is a 544 digit integer (just hold on with me here), color every pixel black for True, and white otherwise, you'd get:

This doesn't end here. If \(k\) is now replaced with another integer containing 291 digits, we get yours truly:

The Tex User Group 2019 conference was held between August 9-11, 2019 at Sheraton Palo Alto Hotel, in Palo Alto, California.

I wanted to attend TUG 2019 for two main reasons - to present my work on the “XeTeX Book Template”, and also to meet my favourite computer scientist, Prof. Donald Knuth. He does not travel much, so, it was one of those rare opportunities for me to meet him in person. His creation of the TeX computer typesetting system, where you can represent any character mathematically, and also be able to program and transform it is beautiful, powerful and the best typesetting software in the world. I have been using TeX extensively for my documentation and presentations over the years.

Day I

I reached the hotel venue only in the afternoon of Friday, August 9, 2019, as I was also visiting Mountain View/San Jose on official work. I quickly checked into the hotel and completed my conference registration formalities. When I entered the hall, Rishi T from STM Document Engineering Private Limited, Thiruvananthapuram was presenting a talk on “Neptune - a proofing framework for LaTeX authors”. His talk was followed by an excellent poetic narration by Pavneet Arora, who happened to be a Vim user, but, also mentioned that he was eager to listen to my talk on XeTeX and GNU Emacs.

After a short break, Shreevatsa R, shared his experiences on trying to understand the TeX source code, and the lessons learnt in the process. It was a very informative, user experience report on the challenges he faced in navigating and learning the TeX code. Petr Sojka, from Masaryk University, Czech Republic, shared his students’ experience in using TeX with a detailed field report. I then proceeded to give my talk on the “XeTeX Book Template” on creating multi-lingual books using GNU Emacs and XeTeX. It was well received by the audience. The final talk of the day was by Jim Hefferon, who analysed different LaTeX group questions from newbies and in StackExchange, and gave a wonderful summary of what newbies want. He is a professor of Mathematics at Saint Michael’s College, and is well-known for his book on Linear Algebra, prepared using LaTeX. It was good to meet him, as he is also a Free Software contributor.

The TUG Annual General Meeting followed with discussions on how to grow the TeX community, the challenges faced, membership fees, financial reports, and plan for the next TeX user group conference.

Day II

The second day of the conference began with Petr Sojka and Ondřej Sojka presenting on “The unreasonable effectiveness of pattern generation”. They discussed the Czech hyphenation patterns along with a pattern generation case study. This talk was followed by Arthur Reutenauer presenting on “Hyphenation patterns in TeX Live and beyond”. David Fuchs, a student who worked with Prof. Donald Knuth on the TeX project in 1978, then presented on “What six orders of magnitude of space-time buys you”, where he discussed the design trade-offs in TeX implementation between olden days and present day hardware.

After a short break, Tom Rokicki, who was also a student at Stanford and worked with Donald Knuth on TeX, gave an excellent presentation on searching and copying text in PDF documents generated by TeX for Type-3 bitmap fonts. This session was followed by Martin Ruckert’s talk on “The design of the HINT file format”, which is intended as a replacement of the DVI or PDF file format for on-screen reading of TeX output. He has also authored a book on the subject - “HINT: The File Format: Reflowable Output for TeX”. Doug McKenna had implemented an interactive iOS math book with his own TeX interpreter library. This allows you to dynamically interact with the typeset document in a PDF-free ebook format, and also export the same. We then took a group photo:

I then had to go to Stanford, so missed the post-lunch sessions, but, returned for the banquet dinner in the evening. I was able to meet and talk with Prof. Donald E. Knuth in person. Here is a memorable photo!

He was given a few gifts at the dinner, and he stood up and thanked everyone and said that “He stood on the shoulders of giants like Isaac Newton and Albert Einstein.”

< />

I had a chance to meet a number of other people who valued the beauty, precision and usefulness of TeX. Douglas Johnson had come to the conference from Savannah, Georgia and is involved in the publishing industry. Rohit Khare, from Google, who is active in the Representational State Transfer (ReST) community shared his experiences with typesetting. Nathaniel Stemen is a software developer at Overleaf, which is used by a number of university students as an online, collaborative LaTeX editor. Joseph Weening, who was also once a student to Prof. Donald Knuth, and is at present a Research Staff member at the Institute for Defense Analyses Center for Communications Research in La Jolla, California (IDA/CCR-L) shared his experiences in working with the TeX project.

Day III

The final day of the event began with Antoine Bossard talking on “A glance at CJK support with XeTeX and LuaTeX”. He is an Associate Professor of the Graduate School of Science, Kanagawa University, Japan. He has been conducting research regarding Japanese characters and their memorisation. This session was followed by a talk by Jaeyoung Choi on “FreeType MF Module 2: Integration of Metafont and TeX-oriented bitmap fonts inside FreeType”. Jennifer Claudio then presented the challenges in improving Hangul to English translation.

After a short break, Rishi T presented “TeXFolio - a framework to typeset XML documents using TeX”. Boris Veytsman then presented the findings on research done at the College of Information and Computer Science, University of Massachusetts, Amherst on “BibTeX-based dataset generation for training citation parsers”. The last talk before lunch was by Didier Verna on “Quickref: A stress test for Texinfo”. He teaches at École Pour l’Informatique et les Techniques Avancées, and is a maintainer of XEmacs, Gnus and BBDB. He also an avid Lisper and one of the organizers of the European Lisp Symposium!

After lunch, Uwe Ziegenhagen demonstrated on using LaTeX to prepare and automate exams. This was followed by a field report by Yusuke Terada, on how they use TeX to develop a digital exam grading system at large scale in Japan. Chris Rowley, from the LaTeX project, then spoke on “Accessibility in the LaTeX kernel - experiments in tagged PDF”. Ross Moore joined remotely for the final session of the day to present on “LaTeX 508 - creating accessible PDFs”. The videos of both of these last two talks are available online.

A number of TeX books were made available for free for the participants, and I grabbed quite a few, including a LaTeX manual written by Leslie Lamport. Overall, it was a wonderful event, and it was nice to meet so many like-minded Free Software people.

A special thanks to Karl Berry, who put in a lot of effort in organizing the conference, but, could not make it to due to a car accident.

The TeX User Group Conference in 2020 is scheduled to be held at my alma mater, Rochester Institute of Technology.