3

For a project i'm working on we need to create a Jenkins node in an AWS environment. For this we will work with the docker image which is available on https://github.com/jenkinsci/docker-ssh-slave The Jenkins node will be responsible for executing AWS commands which can only be done from inside the AWS environment (Not outside).

This Jenkins Master is running in our local environment. We will add the node as a permanent node in the Jenkins master. The Jenkins master is not responsible for setting up the Jenkins Node, since it has no permissions to do this ( this is managed in AWS )

However we would like to monitor the Jenkins Slave inside AWS to see if it it still running succesfully. With other docker container we also expose a health check. If the health check fails, the docker container is treated as unhealthy and is restarted accordingly. Just like we do with the other docker containers, we would like to expose such an health port as well on the Jenkins Slave. If it is not responding anymore AWS will restart the docker container and we have a stable system.

We already explored the Nagios monitoring, but it looks like this is exposing the health in the master and not directly in the slave.

Is it available, or how can we add such functionallity? Could you guide us into the correct direction, plugins which needs to be installed, ...

  • Can we open the default HTTP API of Jenkins in such a Jenkins slave (https://wiki.jenkins.io/display/JENKINS/Remote+access+API)
  • Another approach could be that we add and start NGINX in docker and point the healthcheck to NGINX but this way we are not really checking the healt of the the Jenkins slave.
  • 1
    I would really recommend pursuing actual monitoring further (e.g. Nagios, monit, etc.), since Jenkins is not intended to be a monitoring tool. – jayhendren Jun 19 '18 at 17:51
  • 1
    @jayhendren: I think the question is not clear, we are not trying to monitor our slaves with Jenkins. We are using AWS Cloudwatch as a monitoring tool., But in order to do some things we need an HTTP-endpoint in our (docker) slaves where we can verify if this slave is still up and behaving correctly. At this moment, i have no idea how to do this. So any help is welcome. – Rob Van Pamel Jun 20 '18 at 10:30
0

To me it looks like you are trying to reinvent the wheel. As I understood you would like to restart a container if a health check fails. As you are using docker you should use a orchestration platform in my opinion like Fargate (ECS) or Kubernetes (k8s). I know from the latter that it is possible to define healthchecks, like readiness and health probes. If the health check fails, kubernetes will restart the POD (auto healing).

It could be possible that you would argue that k8s is not an option. If that is the case, you could also have a look to AWS OpsWorks. Although I did not try it myself yet, it is able to perform auto healing as well.

  • we are using ECS for the healthchecks, only problem was that Jenkins slaves do not expose a Health check on their container images. So i was wondering what the best solution would be to add such a healthcheck on a Jenkins Slave: 1) Creating a custom image, and add NGINX which exposes a healthcheck or 2) try to use that Jenkins API in a slave which doesnt seem to work. We used option 1 for now and this is working fine. – Rob Van Pamel Aug 23 at 14:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.