Sundeep Machado

How to enable NVIDIA GPU workloads on k3s cluster

2025-12-17T15:42:00.005+05:30

GPU workloads on k3s

K3s is a highly available, certified Kubernetes distribution designed for production workloads. It can also be used for AI workloads.

By default, k3s nodes do not recognize GPUs. In this article, we will enable k3s to work with a GPU.

Step 1 : Install NVIDIA drivers

# Ubuntu / Debian
sudo apt-get update
sudo apt-get install -y nvidia-driver-535
sudo apt-get install -y nvidia-container-toolkit
sudo systemctl restart containerd

Verify Docker/Containerd GPU access:

nvidia-container-cli info

Step 2 : Install k3s

k3s uses containerd, so we must enable the NVIDIA runtime.

Create config:

sudo mkdir -p /etc/rancher/k3s
sudo nano /etc/rancher/k3s/config.yaml

Inside the config file you need to add the following:

write-kubeconfig-mode: "0644"
container-runtime-endpoint: ""

Now install k3s : (the below command can also be tweaked further, more details on k3s installation page)

curl -sfL https://get.k3s.io | sh -

Restart k3s:

sudo systemctl restart k3s

Step 3 : Install helm

Official Helm repository:

curl -fsSL https://raw.githubusercontent.com/helm/helm/main/scripts/get-helm-3 | bash

Point Helm explicitly to k3s kubeconfig

export KUBECONFIG=/etc/rancher/k3s/k3s.yaml
sudo chmod 644 /etc/rancher/k3s/k3s.yaml

Step 4 : Install NVIDIA GPU Operator

kubectl create namespace gpu-operator
helm install gpu-operator nvidia/gpu-operator \
  --namespace gpu-operator \
  --set driver.enabled=false

After some time, you need to create a test yaml file called cuda-test.yaml

apiVersion: v1
kind: Pod
metadata:
  name: cuda-test
spec:
  restartPolicy: Never
  containers:
    - name: cuda
      image: nvidia/cuda:12.2.0-base-ubuntu22.04
      resources:
        limits:
          nvidia.com/gpu: 1
      command:
        - bash
        - -c
        - |
          echo "=== NVIDIA-SMI OUTPUT ==="
          nvidia-smi
          echo "========================="
          sleep 3600

Use below command to spawn a new test pod:

kubectl apply -f cuda-test.yaml
kubectl get pods

Now if everything is successful, you should see something like this;

 kubectl logs cuda-test

You can now use the node for GPU workloads.

How to deploy Google's latest Embedding Model embeddinggemma-300m on Nvidia Triton Server

2025-10-31T17:31:00.003+05:30

EmbeddingGemma on NVIDIA Triton Server

An embedding model is needed to generate embeddings (vector representations that help an LLM to understand things like text, images etc).

Google recently released EmbeddingGemma-300M (a whooping 300 million parameter model) which has very low requirements to run. We test this assumption in this article.

EmbeddingGemma 300M running on Commodity NVIDIA GPU

Step 1. Download the model from hugging face at your local machine

Please create a folder called "server" and inside that create another folder called model_repository.

We need to download the model from Hugging face. We first need to create an access token by visiting this Hugging face link. We need to give proper rights to the token to download the Google Gemma models:

Grant Gemma model repository access in Hugging face

As we can see above, we have only granted read repos access.

Now, we can download the model by using the below command:

pip3.12 show huggingface_hub
pip3.12 show huggingface_hub_cli

hf download google/embeddinggemma-300m \
  --local-dir model_repository/embeddinggemma-300m/1/embeddinggemma-300m \
  --local-dir-use-symlinks False

Triton Inference server expects the model to be downloaded in a particular format. Please make sure that the files are downloaded properly.

Step 2. Create the model.py file and config.pbtxt file

The config.pbtxt file in NVIDIA Triton Inference Server is a crucial component for defining the configuration of a deployed model. It is a Protocol Buffer Text Format file that describes various aspects of how a model should be loaded, executed, and managed by Triton.
The complete code can be found here
Folder Structure

model_repository/embeddinggemma-300m/1/embeddinggemma-300m/ (empty placeholder — drop the local HF snapshot here)

models/
└── embeddinggemma-300m/
    ├── config.pbtxt
    ├── 1/
    │   ├── model.py
    │   └── embeddinggemma-300m/ (empty placeholder — drop the local HF snapshot here)
Step 3. Create a Dockerfile and build an image
We need to create a custom Dockerfile with following content:

FROM nvcr.io/nvidia/tritonserver:25.05-py3

# GPU PyTorch (CUDA 12.1 wheel bundles its own CUDA, works fine in this image)
# 1) PyTorch nightly with CUDA 12.4 (>= 2.6)
RUN python3.12 -m pip3.12 install --no-cache-dir \
    --index-url https://download.pytorch.org/whl/nightly/cu124 \
    torch --pre

RUN python3.12 -m pip3.12 install --no-cache-dir \
    "transformers>=4.46.1" \
    "safetensors>=0.4.5" \
    numpy \
    "protobuf>=3.20.3,<5.0" \
    "sentencepiece>=0.1.99"


# 2) Transformers that supports gemma3_text + friends
#RUN python3.12 -m pip3.12 install --no-cache-dir \
#    "transformers>=4.46.1" \
#    "safetensors>=0.4.5" \
#    numpy

Now we build this docker file and create a custom image. This dockerfile should be inside the server folder

sudo docker build -t triton-gemma-cpu

Now we run NVIDIA Triton server as a docker container:

sudo docker run --rm --privileged --gpus all  -it   -p 8000:8000 -p 8001:8001 -p 8002:8002   -v $PWD/model_repository:/models   triton-gemma-cpu   tritonserver --model-repository=/models

The above command runs the Triton Inference Server in a Docker container with full GPU access. It maps the standard Triton ports (HTTP, gRPC, Metrics) to your local machine and, most importantly, links your local model_repository folder to the server's model directory, allowing it to serve the models you have stored there.

Now test the model is running by using curl

curl -s http://localhost:8000/v2/models/embeddinggemma-300m

The response will be :

{"name":"embeddinggemma-300m","versions":["1"],"platform":"python","inputs":[{"name":"TEXT","datatype":"BYTES","shape":[-1,-1]},{"name":"MODE","datatype":"BYTES","shape":[-1,1]}],"outputs":[{"name":"EMBEDDINGS","datatype":"FP32","shape":[-1,-1,768]}]}

Sample request:

curl -X POST localhost:8000/v2/models/embeddinggemma-300m/infer -d \
'{
  "inputs": [
    {
      "name": "TEXT",
      "shape": [ 1, 2 ],
      "datatype": "BYTES",
      "data": [ "This is the first document", "This is the second document" ]
    }
  ]
}'

Getting Started with Cloudflare's new AI Crawl Control

2025-09-02T11:50:00.001+05:30

AI Crawl Control From Cloudflare

Cloudflare has recently announced a new feature called AI Crawl Control .

What is AI Crawl Control From Cloudflare?

Cloudflare's AI Crawl Control is a tool that gives website owners granular control over how AI services access their content. It evolved from Cloudflare's AI Audit tool to empower content creators with more detailed insights and agency over which AI crawlers are permitted to access their data, as well as for what purpose.

The solution is designed to address the challenges faced by content creators, who often see their content scraped by AI models without compensation or referral traffic back to their sites.

We have discussed in details on how we can use robots.txt for websites. As a website owner you can specify rules in robots.txt and ask a web crawler like Google Bot what websites it can browse (or index).

Consider the robots.txt file available here -> https://www.sundeepmachado.com/robots.txt

Sample Robots.txt file

User-agent: Mediapartners-Google: This line targets a specific Google bot. Mediapartners-Google is the user agent for Google's AdSense service. It crawls your site's content to serve relevant ads.
Disallow:: The Disallow: directive is empty. This means nothing is disallowed for the AdSense bot. The owner of this website is explicitly giving the AdSense crawler full permission to access the entire site.
User-agent: *: The asterisk * is a wildcard that means this rule applies to all other web crawlers (including Google's main search bot, Bingbot, etc.) that don't have a more specific rule set for them. This also includes all AI agent crawlers like those from OpenAI
Disallow: /search: This tells all crawlers not to crawl any URLs that are in the /search/ directory of the website. This is a common and recommended practice. Internal site search result pages are often considered "thin content" or duplicate content by search engines, so blocking them from being indexed is good for SEO.
Allow: /: This directive explicitly permits crawlers to access the root (/) and all other pages and subdirectories of the site not covered by a Disallow rule. It reinforces that the rest of the site is open for crawling.

Cloudflare AI Crawler Dashboard

As you can see above you can allow/block a particular bot.
As seen above archive_org bot is violating the robots.txt file by accessing the /search url pattern which is mentioned in the fourth point above.

Premium Feature : Pay Per Crawl

Cloudflare's Pay Per Crawl is a system that allows website owners to charge AI companies for crawling their online content.

Using this feature, a site owner can choose to freely allow, completely block, or charge a specific price per request for known AI bots. The system uses the 402 Payment Required HTTP status code to signal the need for payment, with Cloudflare handling the entire transaction process.

How to do a Performance Test for AI Models hosted on NVIDIA infrastructure?

2025-06-28T01:05:00.003+05:30

A performance test is extremely important for an Artificial Intelligence (A.I.) Model just as it is important for an e-commerce website like Amazon.

The NVIDIA Triton server is a gold standard that standardizes AI model deployment and execution across every workload and it is important that it works well for your custom models.

If you/or your company are working with NVIDIA chips, there is a very high probability you are using the Triton server. The Triton server is the secret sauce that helps to get the peak performance out of expensive NVIDIA hardware.

There are some tools provided by NVIDIA as part of it development sdk. You just need to download the correct docker container

1. Triton Model Analyzer - This tool helps to understand the compute and memory requirements for your model. It also uses the Triton Performance Analyzer to test the inference performance of models. (see the next point)

2. Triton Performance Analyzer - This tool only concentrates on the inference performance of models. It helps you to fine tune various model configurations like instance groups etc.

3. Gen-AI-Perf - This tool primarily focuses on Generative AI models.

It currently supports different types of models like Embedding, Ranker and even Vision models.

4. Traditional Tools - The performance testing tools like Apache JMeter and Locust are still widely used.

Getting Started with NVIDIA Triton Server

2024-02-13T23:47:00.001+05:30

NVIDIA Triton Inference Server

What is an Inference Server?

The role of an Inference Server is to accept user input data and pass it to an underlying trained model in the required format and return the results. It is also widely known as a Prediction Server as the results are nothing but predictions (in most cases).

The NVIDIA Triton server is a gold standard that standardizes AI model deployment and execution across every workload and it is important to know how it works internally for your custom or off the shelf models.

The steps to install Triton Server are mentioned in detail on this Triton Github page.

There are few important things to know about Triton Inference Server:

NVIDIA recommends running it as a docker container, so that multiple instances can be run via Kubernetes
You need to install the NVIDIA Container Toolkit especially if you are planning to work with a NVIDIA server class GPU like A100.
You can load new ML models on the fly, however NVIDIA does not recommend this as allowing dynamic updates to the model repository can lead to arbitrary code execution. The best way is to restart Triton each time with a command-line argument of the folder which contains the newly added models.
There is a Prometheus Metrics end-point for GPU, CPU utilization, Memory etc. We can easily send this data to Grafana to create custom dashboards. You can customize the metrics you want


NVIDIA Triton Server running on commodity NVIDIA GeForce GTX 1660 Super

As you can see above, I have successfully deployed the models on the NVIDIA Triton server as discussed on this GitHub page on my Ubuntu Desktop.

Getting Started with Ollama.ai

2023-11-21T12:56:00.007+05:30

Ollama.ai

Ollama.ai is an excellent tool that helps you to run Large Language Models (LLMs) locally on your computer like Llama2.

In this article, I decided to test whether Ollama can work with my consumer grade GPU - MSI GTX Super 1660

My Desktop configuration is as follows:

Ubuntu 22.04.03LTS Linux
24 GB DDR4 RAM
Intel i3-8100 @3.60GHz with 4 Cores
MSI GTX Super 1660 with 6 GB DDR6 RAM

Before we start, we need to make sure that we have the greatest and the latest NVIDIA drivers and NVIDIA CUDA drivers: https://developer.nvidia.com/cuda-downloads. The installation was easy and you need to validate whether your GPU drivers are working by using nvidia-smi command as below:

If you do not have a Graphics Card, Ollama can run in CPU only mode too

nvidia-smi command used to validate whether your GPU has the correct NVIDIA drivers

As you can see above, my NVIDIA graphics card has been detected and is ready to use.

We are going to use NVITOP to measure GPU Performance. It is one of the best free tools to measure GPU performance for free

You can install Ollama fron this official Ollama download link. As of November 2023, you can install Ollama only on a Mac or a Linux machine. Windows support will come soon.

You can now download the llama2 model by entering the following command in your terminal:

ollama pull llama2

Similarly, if you want the Mistral model, you need to enter:

ollama pull mistral

There is a comprehensive list of models that you can use on Ollama Website

Ollama has a REST API to run and manage Machine Learning models. You can call the API as below:

curl http://localhost:11434/api/generate -d '{
  "model": "llama2",
  "prompt":"What is Life?"
}'

As you can see above, we are using the llama2 model.

We can also use Ollama to work with a language model like The Mistral 7B model released by Mistral AI by giving the model parameter as mistral.

Calling Ollama API with Mistral 7 b model:

We will use the following prompt

{
  "model": "mistral",
  "prompt":"What is the capital of the United States of America?",
  "stream":false

}

The response was:

{
    "model": "mistral",
    "created_at": "2023-11-21T07:06:57.825617661Z",
    "response": "The capital city of the United States of America is Washington, D.C.",
    "done": true,
    "context": [
        733,
        16289,
        28793,
        28705,
        1824,
        349,
        272,
        5565,
        302,
        272,
        2969,
        3543,
        302,
        4352,
        733,
        28748,
        16289,
        28793,
        13,
        1014,
        5565,
        2990,
        302,
        272,
        2969,
        3543,
        302,
        4352,
        349,
        5924,
        28725,
        384,
        28723,
        28743,
        28723
    ],
    "total_duration": 3314902170,
    "load_duration": 2814112001,
    "prompt_eval_count": 20,
    "prompt_eval_duration": 152161000,
    "eval_count": 16,
    "eval_duration": 335794000
}

During the test, I had kept nvitop running in another terminal:

NVITOP command output to monitor GPU (Click to Zoom)

As you can see the GPU Memory shot up to almost 5GB. This is because, 7B parameter models generally require at least 8GB of RAM. The actual size of the model was around 4.1GB.

You can similarly download other models and work with Ollama.ai. You can use chat to send your prompts as shown below:

You can invoke the chat interface by entering the following in the terminal:

ollama run mistral

A list of things to remember when deploying a Large Language Model (LLM) on Production

2023-11-09T15:05:00.003+05:30

LLAMA

A small checklist that intends to make your LLM deployment easier.

This checklist is intended to help you get started with deploying your own Model or an open source one like Llama2

1. There is no perfect model that will fit your use case 100%

Although the excitement around Large Language Models (LLM) is justified, very few people actually know in advance which model is the best for them. A very good starting point is looking at a Model Card, if it exists. There are some models that perform some particular tasks better than others. For example, there are thousands of models that cater to Summarization use case. After selecting a particular model, you will need to validate whether it performs particularly well with your in house proprietary data.

2. You don't always need a GPU

A GPU is very useful when you need to work with a very large dataset and need a very high bandwidth for your use-case. The average price of a NVIDIA H100 is $30,000 which is around ₹ 2.5 million Indian rupees and that of NVIDIA A100 is $10000 which is around ₹ 8,33,000 Indian rupees.There are a lot of use cases that don't need a GPU. If you are experimenting with a small use-case, you can get most of a CPU to make your models work. You can easily spin up a CPU heavy virtual machine on a cloud provider.

3. Expect very high response time

The response time of a Large Language Model is very high. There are a lot of factors at play like below:

Parameters: Parameters in large language models (LLMs) are the variables that are learned by the model during training. They represent the weights and biases of the neural network that the LLM is composed of. The number of parameters in an LLM is a measure of its size and complexity. Larger models have more parameters and can learn more complex relationships between words and phrases.

Temperature: Temperature in LLM is a hyper-parameter that controls the randomness of the model's output. A higher temperature will result in more creative and imaginative text, while a lower temperature will result in more accurate and factual text.

K inputs: K inputs in LLM refers to the number of tokens that the model considers when generating the next token. This is a hyper-parameter that can be tuned to control the creativity and diversity of the generated text.

Prompt: A prompt in LLM is a piece of text that is used to guide the model to generate a specific output. Prompts can be simple or complex, and they can include instructions, questions, or any other type of text that provides the model with context.

Nvidia H100

The output of a Large Language Model depends on many factors. It also depends on your infrastructure and system architecture. It is a matter of fact that models running on a GPU have lower inference times than equivalent models running on a CPU.

Note : Depends on underlying Hardware of CPU/GPU

NVIDIA H100 -

4. Use existing tools as much as possible, if that fails accept new ones

There are pretty amazing deployment tools like Langchain, that help in making LLM deployment easier. You can also use your existing Kubernetes clusters for the same. Every cloud vendor is now making it easier to deploy Machine Learning model quickly and efficiently.

5. Secure your LLM - MLSecOps and Model Bias

MLSecOps, or Machine Learning Security Operations, is a discipline that combines cybersecurity practices with ML operations to protect models, data, and the ML infrastructure. MLSecOps is concerned with identifying and mitigating the unique security risks associated with machine learning systems, such as adversarial attacks, data poisoning, and model bias.

Security and Bias Prevention should not never be an afterthought

Model bias refers to when a machine learning model systematically makes errors or exhibits unfair behavior towards certain group. Model abuse refers to the malicious use of AI/ML models to cause harm.

These are real and new problems and the best minds around the world are trying to solve them in innovative ways. Always Learn and Improve.

Free Kubernetes cluster on Oracle Cloud using k3s

2022-05-04T15:27:00.007+05:30

K3S

Oracle Cloud has a very generous forever free tier. I have been using k3s on my Raspberry Pi 4 machines for sometime in my local home network and it is working amazingly well.

I was very keen to deploy k3s for free on a cloud provider as a backup to my local clusters and finally managed to do that recently.

How to deploy k3s on Oracle Cloud ?

Kubernetes k3s cluster on Oracle Forever free tier

We are going to use the forever free tier. As of December 2021, Oracle is giving us below options:

ARM based machines with 24 GB RAM and 4 CPUs cores in total - I have decided to utilize this as 2 separate machines with each having 12 GB RAM, 2 CPUs and 15 GB default volume storage. One machine will be main node for k3s and other will be a worker node
Two AMD machines with 1 Core and 1 GB RAM - I have used one machine as a k3s worker and the other as as MySQL database for high availability for k3s cluster

The installation of k3s is pretty straightforward. It is important to remember that we have taken one node as master and rest as slaves.

Machine Setup

On Master node (we will call it k3s-master)

Ampere ARM instance with 2 Octacore CPU and 12 GB RAM

1. Visit Oracle Cloud create instance and create a VM (virtual machine) with Ubuntu as image and the CPU configuration as ARM and hard disk around 15 GB. This will be k3s-master

2. Now remember and download both the ssh keys as we need them for other machines:


SSH key menu in Oracle Cloud

3. Once the machine is green (shows active in the instances section of Oracle Cloud), use the ssh key to login via the public ip address:

ssh -i ssh -i id.rsa ubuntu@<PUBLIC IP ADDRESS>

Please change the IP address as per your machine. You might need to install Putty if you are using a Windows machine. Change the hostname by entering on the terminal:

hostnamectl set-hostname k3s-master

hostnamectl

On Slave nodes (we will call it k3s-mysql, k3s-amd-node1 and k3s-arm-node2)

1. Create another instance with AMD as CPU type, 10 GB HDD and 1 GB of RAM. This should be within free limits. This will be k3s-mysql (use hostnamectl command as before). Remember to specify the public ssh key you downloaded before in the ssh key section. We need to install mysql on this machine which can be done as per this article - install MySQL on ubuntu. Make sure you create a new user and a database. Give admin rights to this user to that database only in MySQL You need to note the private IP of this machine.

2. Create another instance with AMPERE ARM as CPU type, 2 OCTACORE CPU, 10 GB HDD and 12 GB of RAM. This should be within free limits. This will be k3s-arm-node2 (use hostnamectl command as before). Remember to specify the public ssh key you downloaded before in the ssh key section. Note the private IP address of all machines so that we can use them and also change their hostnames

Installing k3s

1. Login to every machine and make sure you enter the following command:

sudo iptables -F

This commands flushes the iptables rules.

2. Now in k3s-master (use ssh to log in), we will install k3s server in high availability mode using the below command:

curl -sfL https://get.k3s.io | sh -s - server \
  --token=SECRET \
  --datastore-endpoint="mysql://username:password@tcp(hostname:3306)/database-name"

(Please replace the username, password, hostname - private IP of k3s-mysql and database-name as per your configuration)

3. Note the token string (K3S_TOKEN) which you get on k3s-master by using following command:

cat /var/lib/rancher/k3s/server/token

This will be used by worker nodes to join the k3s servers

4. Now log in to each worker node and enter following command:

K3S_TOKEN=SECRET k3s agent --server https://fixed-registration-address:6443

fixed-registration-address ---> k3s-master private IP address

SECRET ---> K3S_TOKEN from step 3.

That's it. Enjoy your free k3s (Kubernetes) cluster on Oracle Cloud.

US Elections 2020 - The best way to keep track of what is happening

2020-11-04T17:18:00.002+05:30

US Elections 2020 is too close to call

The US Elections 2020 is one of the most widely watched event in the world.

We look at the ways the Tech / Media Companies are trying to make sense of the data.

Google

A simple search for US Elections shows this Minimalist Widget

Google is leading the way in showing a minimalist widget that gives you accurate data from Associate Press. This works quite well on Mobile devices too. You can narrow down further to get specific data about "Who is winning the Senate" etc.

The Guardian Website

The Guardian Website has created a Live Blog for US Elections

The Guardian website is one of the most technically advanced news website. It is known for having a right blend of tech and user friendly aesthetics.

Check for a particular State in real time using data from Associated Press

The New York Times

The New York Times US Election 2020 Website is the advanced version of the Guardian website.

NYT Homepage for US Elections 2020

The only issue is the irritating pop-up that comes up after you exhaust reading the 5 article limit imposed. (you can by pass it by surfing the website in incognito mode)

The Washington Post

The Washington Post US Elections 2020 Homepage

The Washington post too has a very good minimalist website that you can use to keep track of what is happening.

Microsoft Clarity - An easy to use User Behavior Analytics Tool

2020-10-30T11:06:00.000+05:30

Microsoft Clarity is a free service that helps users to understand how users are interacting with their websites.

The services offers the ability to record user sessions and also generates a Heat Map for clicks.

Microsoft Clarity Heatmap feature

How to Setup Microsoft Clarity?

You ca use your existing Microsoft, Facebook or Google Account to start using Microsoft Clarity. You need to visit the official Microsoft Clarity website to get started.

Creating a new project in Microsoft Clarity

After you log in, you need to setup a new Project in Clarity as above. After that you need to add tracking Code to your website. This needs to go inside the head tag of your website HTML code.

Adding a tracking code for Microsoft clarity

It might take up to 2 hours for data to show up on your Microsoft Clarity Dashboard after the tracking code is inserted on your websites.

Microsoft Clarity and Google Analytics Integration

Microsoft Clarity also has support for Google Analytics, this creates a Clarity user session link in your Google Analytics Dashboard.

Microsoft Clarity does not replace your existing analytics solution like Google Analytics . It is meant to help you understand more about your users behavior. It is a great place to start for a new user who wants to have basic understanding of website users. There is a competing free service from Yandex Metrica too.

Using Gitlab for backing up data to AWS S3 and Google Cloud Storage

2020-02-18T17:24:00.003+05:30

Gitlab has a generous free tier for CICD called "Gitlab Pipelines", that can be used to store build artifacts (or anything) in AWS S3 and Google Cloud Storage for free.

Grab the code here: GitlabToCloud.

How to use Gitlab to store code artifacts like jars or zip files?

You will need a Gitlab account. You will also need an AWS account that has suitable access keys, (Secret key and Access key )to write to a bucket in AWS S3.
Similarly we also need to have proper service accounts for writing data in a bucket for Google Cloud.
You will also need to setup Gitlab environment variables. The variables are as follows:

$AWS_ACCESS - AWS access key
$AWS_SECRET - AWS secret key
$DAYS - Set it to 1 days
$GCP_PROJECT_ID - The Google Project id
$GCP_SERVICE_KEY - The GCP service account created in step 2 above. Just copy all the file contents

Grab the code here: GitlabToCloud.

As you can see, we are using the Google Cloud image as it has python and gsutils already installed. We are installing aws-cli using pip so that we can easily copy the required files. All the files are deleted for security reasons after execution.

Using AWS lambda as a Proxy for any service

2020-02-06T12:04:00.001+05:30

Amazon Web Services has a generous free tier for Lambda which involves 1 million free invocations.

With the release of HTTP API for API Gateway, you can use Lambda as a proxy for any service.

The source code is available here -> AWS Lambda Proxy using Nodejs

Why would you use Lambda as a Proxy?

Security - We don't want to expose the API end-point directly to the internet
Resilience - If the external service goes down, we can use lambda to re-direct to an another instance of the API.

How to use AWS Lambda as a Proxy?

AWS Lambda Proxy Architecture

You start by creating a new lambda function (we will call it the proxyJS). This will be the end-point where user submits the form.

API Gateway Trigger for Proxy Lambda (Click to enlarge)
Now we can call any external end-point. But in order to de-couple our architecture, we will use an AWS SNS topic that will get data from the Proxy lambda.

SNS Trigger for AWS Lambda

This topic will be subscribed by another Lambda function called externalJS that will call an external service with the topic data
In the above architecture, we are calling an external Google Cloud HTTP function using Nodejs using the above lambda

Getting Started with Fortio for Load testing

2018-08-01T22:20:00.002+05:30

We look at Fortio, which is a powerful load testing tool which is free and powered by Google's Go language.

It is a rightful successor to Vegeta which is favourite among Go developers.

Updated : 16 November 2023

Installing Fortio

Install go (golang 1.8 or later)
go get istio.io/fortio
you can now run fortio (from your gopath bin/ directory)

How to run a load test using Fortio?

fortio load -qps 100 -t 30s http://example.com/api

This command will perform a load test at 100 queries per second for 30 seconds against the specified URL (http://example.com/api). You can adjust the options (-qps for queries per second, -t for duration) based on your requirements.

Docker

The following docker commands are useful when working with Fortio

docker run -p 8080:8080 -p 8079:8079 fortio/fortio server & # For the server

This runs Fortio in GUI mode (this open a web interface at http://localhost:8080/fortio)

As you can see, you can specify the percentiles for reporting, the TPS and the duration using the web interface. However, you are only supposed to use the GUI mode for developing your scripts

docker run fortio/fortio load http://www.google.com/ # For a test run

This command runs fortio without the GUI mode and is recommended for a test

The command-line gives you more options but it is not as powerful as a Apache JMeter or Gatling . It does not support scenario based load testing

Things I like:

A simple web interface
Ability to share report and graphs

Things I don't like:

No support for scripts
No support for scenario bases load testing

Fortio is best suited for load testing where developers want to test their code locally on their desktop computers. It can also be used for simulating spikes.

New User SignUp with Welcome Email using Firebase FireStore, Cloud Functions and Mailgun

2018-03-24T00:50:00.000+05:30

Cloud FireStore and Cloud Functions can be easily used to save user details when a new user creates an account. An email can also be sent easily using an external service like Mailgun.

The basic idea is as follows:

1. User Signup and entry in Cloud FireStore

The above code is triggered when a new user signs up for the first time. It should be triggered only when the Firebase Auth is successful.

In the above code snippet, id is nothing but the unique id returned by Firebase Auth. The field "name" is entered by the user during signup.

2. Cloud Function to trigger Email

It is important to note that we need to have proper Mailgun API credentials. Also mailgun.js should also be installed in the directory of the cloud functions.

The above code works for all types of login, but the user should have a valid email id in the Cloud FireStore.

Should you buy a FireTV or a Chromecast?

2017-07-25T22:21:00.000+05:30

Amazon FireTV 2 and Google Chromecast 2 are one of the best living room devices available. (other than Roku and Apple TV).

So which one should you buy?

The Price

Google Chromecast 2 costs $35.00 dollars Amazon Fire TV 2 costs about $39.99 dollars. Google Chromecast is the clear winner here. However, on special days like Amazon Prime Days you can get a FireTV 2 for about $20 dollars.

The Apps and Software

The Google Home App lets you configure your Chromecast. It does nothing more. The Amazon FireTV app lets you control your FireTV (like a remote).

In terms of easy of use Google Chromecast is the clear winner hands down. You can cast your videos on YouTube or on Netflix using a cast button. All you need is an Android device or an iOS device with apps that support casting like YouTube.

FireTV targets people who prefer the native TV experience. You can install new apps and also side-load additional android apps (install apps not available in Amazon FireTV Appstore).

The clear winner in terms of features is FireTV. You can also use your TV remote to control FireTV.

Verdict

Google Chromecast makes total sense if you just want to cast casual videos and value simplicity. Amazon FireTV gives you a holistic TV experience that is totally worth the price given you buy it on Amazon Prime day Sale.

How to find your lost or misplaced Android device?

2017-07-05T10:06:00.002+05:30

Google's Find My device is a free Android app that helps users to locate a missing android device.

You just need to install the app and then you can locate your device using your Google account.

Steps to find your missing Android device:

Install the Find My Device app on your Android device (it is very important as the app helps in locating the device)
Visit the Find My Phone website, you will now see the list of devices as below after you log in to your Google account:
Click on the device you want to locate.

Additional Options:

You can have additional options after you locate your device like locking the device, erasing all data etc. as seen above.

Get free 1000 email accounts for your domain

2017-05-21T11:14:00.000+05:30

Yandex.Mail for Domains is a free service provided by Yandex. After verifying your domain name ownership, you will get 1000 email accounts with 10 GB space for free.

Steps to get free email accounts via Yandex.Mail for Domains

Create a Yandex account
Verifying your domain ownership

In order to verify the domain you need to add the following domain Nameservers:

Primary DNS server — “dns1.yandex.net.”
Secondary DNS server — “dns2.yandex.net.”

After verification, you can create upto 1000 email accounts. You can also use the optional Domain Management services too.

How to upgrade your website or blog from http to https for free

2017-05-16T20:00:00.000+05:30

Cloudfare offers a free SSL certificate that enables a website publisher to host content using the https protocol on a website.

It also allows a blogger blog which has a custom domain to upgrade to https.

What are the advantages of https?

Websites are more secure
Google assigns a higher ranking to https websites
Websites are faster

How to install Cloudfare certificates?
In order to install the Cloudfare SSL certificate so that all content is served via https, a website owner needs to change the DNS nameservers to that provided by Cloudfare as below:

The best thing is that existing MX, A, TXT records are backed up and nothing gets deleted.

You then need to enable SSL in the Crypto tab in Cloudflare console:

As soon as you enable this option, the website will be served via HTTPS protocol. Users will see a lock option in the address bar as below:

You can easily force all mixed content to be served via https using the below setting:

The advantage is that if a website owner has uploaded an image which was served by http and is now available on https, Cloudflare detects this and automatically serves such content via https.

How to replay user actions on your website?

2017-05-08T19:37:00.000+05:30

Yandex Metrica is a free Analytics service similar to Google Analytics.

It has a cool user session replay feature that allows to see how users are using a particular website. It basically records users sessions and then replays it.

It gives a detail overview of your users and the duration of each session. To replay you just need to hit the play button.

Metrica also supports advance reporting options as you can filter users based on particular criteria. Example: You can filter users visiting the website using Chrome Browser and who are visiting from United States of America.

Opera VPN - Free VPN app for Android and iOS devices

2017-04-30T19:36:00.001+05:30

Opera VPN is a free VPN by Opera for android and iOS devices.

It has a lot of cool features that are not available in most commercial VPN service providers.

What are the different features of Opera VPN?
1. Hide your IP address and get a different IP address from a particular country

Opera VPN allows users to select a region particular region like United States and Canada.

2. Block irritating ads and trackers on websites

Advertisers assign an advertising id to users and this helps them to track user activity and know how they are using apps and website.

3. Test Wifi Settings

Opera VPN can also test any wifi settings to check security loopholes .

After Fake news, Facebook introduces Vivo fake ads on Instagram

2017-04-27T20:11:00.000+05:30

Facebook has done it again.

After fake news, Facebook has started showing fake ads on its Instagram service.

An iPhone 5s for Rs 499 is 7.78$. This advertisment is too good to be true. After clicking on the advertisement, the website opened was a similar to Amazon as seen below.

A domain lookup showed the real owner of the website as below:

The advertiser has smartly masked the url with Vivo which is a mobile phone maker.

Facebook chat bot allows anonymous people to ask and answer questions

2017-04-26T23:00:00.001+05:30

The Facebook messenger bot platform has a good number of useful bots.
A new bot called Give or Take allows users to ask and answer questions

This Facebook bot is very fast and very engaging. You need to install the facebook messenger app on your device for it to work on your phone. Once installed visit the bot at GiveOrTakeChatbot .

How to run ad experiments using Google Adsense

2017-04-26T05:30:00.000+05:30

Google Adsense allows content publishers to experiment with different ad formats and ad categories.

You can find the Experiments feature under the Optimisation Tab.

There are two types of experiments:

Ad unit settings - This experiment allows the user to change the the ad type or ad style and see what works best.
Allow and block experiment - This experiment allows the user to work with different categories and op-outs

The idea is to identify which ad formats are working for a particular website and how website owners can maximise their revenue.

Google Chrome to have new ad-blocking features soon

2017-04-24T21:32:00.000+05:30

Google Chrome is all set to introduce new ad-blocking features that will be enabled by default.

Google Chrome is the world's most used Web Browser as per W3Schools. As a result it has monopoly in deciding what ads it wants to display to the users.

The ad blocking market is growing at a tremendous space. About 15 % mobile users have ad-block software installed on their devices.

Adblock Plus is the number one ad-blocker available to the users which helps users to block advertisements on websites like YouTube. As a result companies like Google who own sites like YouTube end up losing a lot of money as users don't see ads. By introducing ad-blocking features, Google ensures that it can show ads on websites the company owns or deems fit.

Google also allows website owners to earn money using its Google Adsense program to which there is no fee.

Google Analytics Chrome plugins : How to use them?

2017-04-23T10:20:00.001+05:30

Google Analytics team has released two free Chrome Plugins:

As soon as you install the plugins, you will see them reside next to Google Chrome address bar.

Google Analytics Debugger

This Google Analytics debugger plugin helps to debug whether Google Analytics is working correctly on a page.

You can easily debug a webpage by enabling the plugin and clicking on the console tab in Google Chrome.

Google Page Analytics Chrome Extension

This extension gives an overview how web site visitors are interacting with the website by visualising the data collected from Google Analytics.