mitzen

When i was building my project for a docker container i bump into this problem "Unable to start Kestrel. System.AggregateException: One or more errors occurred. (Unable to load DLL 'libuv': The specified module or one of its dependencies could not be found." I just use install the nuget package and it works Install-Package Libuv -Version 1.10.0

A typical unit test for Angular looks like this :- We need to understand a) A component or fixture represent our angular component which might include a single or more markups. b) DebugElement are basically specific html markup like span or div that we would like to find after it is our component rendered. So, in our test it is important to know what we've already initialized our component using code below :- Next we get an instance of our component and check if it contains certain markups as shown below :-

Oh my, surprising enuff, I had this happened to me, write. Unable to find module @angular-devkit/core. npm install @angular-devkit/core --save-dev

Facebook Bot Concept  What we are doing here is to serve up an nodejs app which handles messages that users or customer post to facebook. This gives us an opportunity to add various service like Luis or Wit.Ai to handle NLP. Surprising Wit.Ai is very similiar to Luis. Getting started with sample code  Downloading sample app with Facebook messenger test drive. https://developers.facebook.com/docs/messenger-platform/getting-started/test-drive/?testdrive_messengerbot=true Follow the instruction here. Running it  Running using 2 different processes. Process #1. Kicks it off using "node app.js" - default port is 5000 but you can always  change it. Process #2 (tunneling to Facebook) Runs the following command -  lt --port 5060 Testing Open your messenger and start sending messages to it. You can see your local (command prompt) node app displaying text messages from the user. https://developers.facebook.com/apps/

You can install the NLog plugin to logging to a Fluentd node.  Fluentd is a unified logging layer and if you're wondering if we're talking about the same logger, check it out here . There is a difference between fluentd and fluentbit. Fluentd is targeted for servers with larger processing capacity while fluentbt is for IOT devices with small memory footprint. Getting your project ready with the following nuget Install-Package NLog.Targets.Fluentd -Version 1.0.4 Install-Package NLog -Version 2.1.0 Install-Package System.Configuration.ConfigurationManager -Version 4.4.1 Setting up Fluentd Docker This is the docker command to setup Fluentd as an endpoint to gather all your log. docker run -d -p 24224:24224 -p 24224:24224/udp -v /data:/fluentd/log fluent/fluentd This will start your docker container with fluentd. Then we will use the following code to write to docker container :- You can see your log as shown below when you connect to your docker co

We so used to building REST or micro service. Now it makes sense to build better REST based services. Better documentation with  Swashbuckle.AspNetCore It is always good to have SwaggerUI in your system as it gives you auto-generated, nicely layout way of documenting and interacting with your REST service. Install-Package Swashbuckle.AspNetCore Next, configure your startup.cs Remember to browse to your swagger UI by :- http://localhost:5050/swagger/

Here is the code to download MNIST dataset. Just run python download_mnist.py Next you would like to try out logistic regression with this dataset.

Surprisingly deploying to a cluster is pretty straight forward. To run or deploy your application to cluster, issue the following command :- kubectl run --kubeconfig connect_config.txt connectcd --image=kepung/dotnetapp-prod --port=5050 The only different is --kubeconfig - which is a file storing your cluster info including access. Next, to ensure, we expose our port with a service, using command below :- kubectl --kubeconfig connect_config.txt expose deployment/connectcd --type="LoadBalancer" --port=5050 --name=connectcd-http That's pretty much it. To get the service url for your pod, you can use the following command :- kubectl --kubeconfig connect_config.txt get service connectcd-http -o yaml You will see the following output and it contains the DNS of the newly deployed service. Please wait for a few seconds for AWS DNS to have it registered in there.

When working with Microsoft Bot, you there are a few gotchas when working with adaptive cards. Nuget package You need to install "Microsoft.AdaptiveCards" nuget package SubmitAction  If you have data in a property called  "DataJson", it needs to be in Json format, otherwise your layout will not be rendered.

https://docs.microsoft.com/en-us/dotnet/csharp/whats-new/csharp-7#discards

The code sample below gives

Here is a good example of how you can create and evolve your simple fast forward network for a simple classification problem

If outcome is in a binary form, e.g true or false, or 1 or 0 then use sigmoid Otherwise use softmax where outcome can be more than 2 possible states.

CNTK is suppose to be a large scale corporate development of ML. You can call it via C#. Similar to tensorflow, it is designed as a graph based ML development framework. A couple of interesting things i find out Installation CNTK  To install CNTK on Windows. If you're installing for different platform, you can refer to documentation here . pip install https://cntk.ai/PythonWheel/CPU-Only/cntk-2.4-cp35-cp35m-win_amd64.whl Get started  The best place to get started is following instruction here . Using CNTK with C# Yeap, you can use CNTK with C#. Light RNN https://github.com/Microsoft/CNTK/tree/master/Examples/Text/LightRNN Tutorials https://cntk.ai/pythondocs/CNTK_200_GuidedTour.html?highlight=cross_entropy_with_softmax Docker image  You can find the docker image here. docker pull microsoft/cntk

Microsoft's guidelines for building REST based services can be found here . 

When you change individual rows in primeng datatable, datatable will not be refreshed automatically,  for example this .rows[i][ 'gender' ] = 'Female' In order to refresh table, you need to do a manual, shallow copy using code below :- this .rows[i][ 'gender' ] = 'Female' this .rows = [...this.rows]; // refresh manually here ..

Here are some common functions to discover properties and values of an object using properties. a) Object.keys - Get all the keys of an object b) Object.values c) Object.getOwnPropertyDescriptors The following code snippet demonstrate how to work with these methods.

There are 3 different types of bindings in Angular a) Source to View bindings The following syntax are valid  {{expression}} [target]="expression" bind-target="expression" b) View to source binginds The following syntax are valid  (target)="statement" on-target="statement" c) Two way bindings The following syntax are valid  [(target)]="expression" bindon-target="expression"

In order to push to Amazon's ECS registry you need to setup and configure your account. First of all you need to setup aws and docker. Goto AWS console to create a repository and follow the instructions. If you needed a sample project, please clone one here . To summarize, the steps, do the following. (i assume you created a repository called "jeremy". Get a docker login command (setup your credential) by running the following command :- aws ecr get-login --no-include-email --region ap-southeast-2 Then build your image docker build -t dotnetapp-prod . docker tag dotnetapp-prod 312611170176.dkr.ecr.ap-southeast-2.amazonaws.com/jeremy:dotnetapp-prod And push your image to Amazon docker push 312611170176.dkr.ecr.ap-southeast-2.amazonaws.com/jeremy:dotnetapp-prod To browse your ECS repository, please go here . Possible error message :- Can't push image to Amazon ECR - fails with “no basic auth credentials” -

I had this situation whereby kubernetes pods keep on restarting even if I tried deleting it. As soon as it gets deleted, two new instance get created again. So i tried looking for what causes it. For my case, it is the replicationcontroller. kubectl get replicationcontroller kubectl delete replicationcontroller "connectcd" And this solves it! :)

This is a simple guide how to deploy .net core docker image to kubernetes (minikube). You need to setup your minikube, docker. Once you have done that, we're ready to go. Summary of tasks :- a) setup Docker image b) push to Docker registry c) run your application on Kubernetes and scale it 1. Setup Docker image Our .netcore service exposes /hello/index url which returns "Hello World!". The complete source code is available here . Once you have clone the git repo, we just need to build our docker image. This is a multistage docker image which means it uses .netcore sdk for building and then deploy built binaries to a .netcore production image. Go to ConnectCD-NetCoreWebApi, then run the following command :- docker build -t dotnetapp-prod . If you want to test your application locally (using DOCKER), docker run -d -p 5050:5050 dotnetapp-prod 2. Push to Docker registry Use the following command to push your built image to Docke

I am doing this because well, i needed to use kubernetes to run my docker image. It's a harder to setup a local docker image, so this is easier. First you need to create a free user account in docker . Login to docker and provide your username and password docker login Next tag your image (kepung/my_image) where kepung is the name of my user. docker tag my_image kepung/my_image Push it $ docker push kepung/my_image To see your docker images on the web, go to https://cloud.docker.com/swarm/kepung/repository/docker/kepung/dotnetapp-prod/general All rite, some FAQ. If you rebuild your docker image and tag it say "hello-world", existing image gets overwritten If you push the image with the same tag name, your image on the web, gets over-written too. Pretty straight forward.

This error comes up when you try to run a .net core application without specifying target build environment output. You need to specify the "-r linux-x64" parameter and that will generate a list of all the assemblies required to run the dotnet core app in target environment. RUN dotnet publish -c Release -r linux-x64 -o out

I just love this docker approach which use sdk image to build and then deploy to a prod in a single file, which looks like code below :- There are more to it to Dockerfile COPY command than meets the eye.

Here we would like to figure out how to work with different HTTP request methods - A complete list of methods available are :- a) HEAD - same as GET, except it does not accept payload b) GET - access a resource c) POST - submit data with the intention of modifying a resource or adding to a resource d) PUT - modify a resource e) DELETE - remove a resource f) TRACE - Echo back the received resource forms server. g) OPTIONS - Returns what the HTTP server supports. For health check for instance to check if a server is online, I think this would be the method to use. Not because it is light or doesn't have a payload but it makes sense. P.S. There is no semantic usage at the moment from HTTP governing body. h) CONNECT - Convert request connection to a transparent TCP/IP tunnel. Not supported i) PATCH - partial response to a resources request. CONNECT and TRACE are the only options not supported by .Net core.

Say you have a bunch of css and javascript files which you need to deploy as offline asset. Meaning the target server does not have access to internet. What do you do? You can change your ".angular-cli.json" file and under the section "styles", update path to your assets in src folder. Yeap, it starts looking in your src folder and you can tell I have a folder called " assets ". "index" : "index.html" , "main" : "main.ts" , "polyfills" : "polyfills.ts" , "test" : "test.ts" , "tsconfig" : "tsconfig.app.json" , "testTsconfig" : "tsconfig.spec.json" , "prefix" : "app" , "styles" : [ "styles.css" , "assets/bootstrap.css" ], "scripts" : [ "assets/jquery-3.2.1.slim.min.js" , "

Rest properties supported by Typescript are shown below :- Spread example Shallow Copy object

I hit a problem of query timeout. This is a freaking old problem but just that we need to understand differences between connect timeout and query time out. If you don't you find yourself setting a connection string and getting exception over and over again. The main different is message your get as shown here :- Connection time out and its exception message. System.Data.SqlClient.SqlException: Connection Timeout Expired. The timeout  period elapsed while … To solve this you can change the connection string which is pretty straight forward;  'Connection timeout=3600' Query timeout and its exception message System.Data.SqlClient.SqlException: Timeout expired. The timeout period elapsed prior … With this, you can use Command.Timeout or if it is a SQL statement execute statement, you can pass in the connection timeout value.

The easier way is to do this ng build --prod Copy everything from your folder called 'dist' over. If your target host machine are not able to access CDNs for javascript or css, you can copy and save into a style.css and run this command, which essentially will generate javascripts and css files. 

lets say your have this as your Controller method, accepting POST request When dealing with large request object, you can never be sure where you might have issue. In this case, I have an issue converting an empty string to int. Yes, it is very important to have Request model to have int? as the default binding variables. But anyways, working as a team you will never know what really happens. The request always null.

Why don't we use catch when calling http post construct? For example, this . http . post ( APPLICATION_HOST + '/employee/save' , payload , { headers : headersJson }) . catch ( err => { console.log(err) ; } Http Post is lazily instantiated with subscribe method call. That's why the code below is better :-

Such as shocker to get this after relocating my folder. What the heck is going on, I didn't even change my code? It is pretty clear that we have a module issue, which can simply be fixed using npm install @angular-devkit/core --save-dev