AWS Log insights stdout F

 Make sure you replace the following lines:

• Parser              docker

With the following:

• multiline.parser    docker, cri

This is due Kubernetes migrating from docker to containerd.

Kubernetes pods stuck in terminating state

 First try to do the following:

kubectl delete pod -n <namespace>  <pod name> --force --grace-period=0

If that does not work then edit the pod, like:

kubectl edit pod -n <namespace>  <pod name>

And remove the finalizers section and save

finalizers:

- <something here>

Kubernetes Grafana Dashboards

 I was trying to install Prometheus and Grafana in one of my Kube clusters. So I download Prometheus and Grafana helm repos, to my surprise most of the predefined dashboards that I found were not working, so I spent hours figuring out why; there were some missing metrics and components not installed by default. At the end, I was able to make some of those work.

For days, I thought that there was not a prebuilt Grafana/Prometheus dashboard solution until I took a look into Rancher, in their documentation they make a reference to Kube-Prometheus-Stack.  Not sure why I did not find this helm chart before, it already contains a full set of Grafana dashboards that will help you monitoring Kubernetes (pods, workloads, namespaces, etc). 

Here is the link to the Helm Chart.

User defined Java Class getRow null

 I have been working with Pentaho Kettle, and I needed to add some scripts, I decided to use Java since you can import some extra libraries, which is not allowed on Javascript component (or I haven't found how). 

I read the documentation and I implemented the processRows method, while testing I realized that getRow method was returning 'null' every time, at the beginning I thought it was related to the input type so I tried a couple, json, csv, excel, etc, and I always had the same result.

What was the solution? I was missing `putRow(...)` line, I am not sure what internally does but after adding it then getRow() started sending objects instead of null. I mean it makes kind of sense since if you miss putRow then whatever you do inside this component will not propagate further, I just wish it was documented somewhere.

So your code should look like this:

public boolean processRow(StepMetaInterface smi, StepDataInterface sdi) throws KettleException

{

// you need this

Object[] r = getRow();

// this

r = createOutputRow(r, data.outputRowMeta.size());

// and this to send the row on to the next step.

putRow(data.outputRowMeta, r);



return true;

}

jenkins-log.txt Permission denied

I started getting the error below, after recreating Jenkins setup thru helm charts. At the beginning I thought I had missed to install a plugin, or installed one that corrupted Jenkins pipelines. So after starting from scratch I could not find why it was working before and then got broken.

sh: 1: cannot create /home/jenkins/agent/workspace/myproject/durable-48fa5881/jenkins-log.txt: Permission denied

So after reading a thread on Jenkins blog; I realized that "Jenkins" user is being used by Jenkins, and I was using "myuser" in my pod. So, I changed everything in my Dockerfile to match Jenkins slave image, like the following:

RUN groupadd --gid 10000 jenkins && useradd --gid 10000 --uid 10000 jenkins --shell /bin/bash
RUN mkdir /home/jenkins && chown -R jenkins:jenkins /home/jenkins
USER jenkins
WORKDIR /home/jenkins

After that, the permission denied error went away. I am still wondering how this was working before...

sbt-native-packager /opt/docker permission denied

I was getting all sorts of issues when trying to create files or directories below "/opt/docker" path. I tried adding:

import com.typesafe.sbt.packager.docker.DockerChmodType
dockerChmodType := DockerChmodType.UserGroupWriteExecute

It did not work, I was still getting errors, permission denied, while trying to write PID file to "/opt/docker/PID_FILE".

The solution is adding the following:

javaOptions in Universal ++= Seq(
  // Since play uses separate pidfile we have to provide it with a proper path  // name of the pid file must be play.pid  s"-Dpidfile.path=/var/run/play.pid")

Basically indicating Play framework to write PID file in another location, then:

dockerCommands ++= Seq(
  ExecCmd("RUN", "mkdir", "-p", s"/var/run/")
)

It would create another directory. Then, execute:

sbt dist && sbt docker:publishLocal

Finally you can write the following in a bash file:

dir=$(pwd)
docker run -v $dir/my_play_project:/var/run -v \ 

$dir/my_play_project/logs:/opt/docker/logs -p 9000:9000 \ 

 my_play_project:1.0

NullPointer Exception while testing with Scala Futures...

So I was trying to unit testing Scala Futures, and I was getting NullPointerException out of nowhere. I was frustrated since the stacktrace didn't much more information, not a minimal clue of what was missing. It was until I payed attention to some other examples when I realized that I was missing something important:

import ExecutionContext.Implicits.global

I do not know why it wasn't detected at compile time, or why the exception was only coming as NullPointerException without any other hint. My test is something like this:

"Some test" should {
    "execute this" in {   
      val client = mock[AClass]
      val host = new BClass(client) 

      doReturn(Future {x}).when(client).aMethod(any[String], any[String]) // This was throwing a NPE

      val response = host.anyMethod(cI, mA, bL)
      Await.result(response, Duration(30, SECONDS)) 

      there was one(client). aMethod(any[String], any[String])
    }
  }