I have built a DCOS local universe and installed it into a cluster behind a firewall - there is no internet access to the cluster. One of the packages installed in the universe is Flink. I have installed DCOS using the cluster_docker_registry_url variable pointing at a local Docker registry which has a very small number of packages on it; it is not a mirror of the main Docker Hub.
When I try to install the Flink package into DCOS, I get 404 errors in the Mesos logs relating to missing docker images that I assume the package tries to download from the local Docker registry. The Flink cluster fails to start.
What Docker images does the Flink package try to download? I thought the build process of a local universe pulled all dependencies down when it is built, so there should be no external dependencies once it's built? What do I need to do to be able to install DCOS when there is no internet access?
That depends on the scala version you are using:
scala 2.10: mesosphere/dcos-flink:1.2.0-1.4
scala 2.11: mesosphere/dcos-flink-2-11:1.2.0-1.4
See here
Furthermore, it requires
openjdk:8-jre ,see here
For more details feel free to refer to the universe specification for the Apache Flink service (or ping me directly):
https://github.com/mesosphere/universe/blob/version-3.x/repo/packages/F/flink/1/
Related
I currently have a Windows GitLab runner installed on a Windows server with Node v15.14.0 installed, but when I attempt to run the single job in my pipeline, it appears to run with Node v17.4.0.
What might be the cause for the discrepancy?
Screenshot of the Node version
My best guess is that the GitLab Runner host machine has multiple versions of Node installed.
I would recommend creating another GitLab Runner using the Docker executor, instead of the Shell executor, and specify an appropriate docker image (e.g. - node:15.14.0) to use for the job.
I downloaded Confluent Platform in my local windows machine & tried to start zookeeper, but it is giving me below error:
c:\confluent>.\bin\windows\zookeeper-server-start.bat .\etc\kafka\zookeeper.prop
erties
Classpath is empty. Please build the project first e.g. by running 'gradlew jarA
ll'
Confluent does not test their products on Windows, last I heard.
The recommendation is to install WSL or use the Confluent Docker containers.
I'm using Jenkins with pipelines on a mac-mini. All builds are working fine with docker agents (backend, frontend, android app, etc)
The only thing I haven't been able to achieve is to use my own mac-mini as build-agent/slave for the IOS app (I need to build on OSX). Jenkins itself runs through docker as well, so I would need to connect to the host (the OS of the mac-mini) and use that as an agent...
I know one option would be to install jenkins instead of using docker, but I would prefer to keep Jenkins running in a docker container.
Does someone has experience with this or knows any good documentation on how to set this up?
Go to Manage Jenkins > Manage Nodes > New Node.
Configure a node.
Go to the list of nodes.
Select your newly configured node. It should be offline at this moment.
Run the java command displayed on the interface on your host machine.
Your Host machine is now a slave.
The docs for the Workload Scheduler for Node.js says:
"Important: Before running a database step, download and install the
JDBC database client driver on the agent where you want to run the
step. Specify the client jar file path in the JDBC jar class path."
How can I download and install the necessary JAR files to the agent? I see from this question that they should be installed at /home/wauser/utils, but I cannot figure out how to access the agent to install.
I tried an FTP step to move the file to the agent, but it was also unsuccessful.
From your description I assume your are trying to run SQL steps on the xx_CLOUD agent.
Where is the database running and which type of database is it? is it on Bluemix or somewhere else?
Currently the best way to run SQL steps is to use a workload scheduler agent installed on a VM or a docker image, so that you can easily install the jdbc jar files.
I need to create a Build Server in CentOS 6.4 Minimal I sucessfully installed:
Java compiler (OpenJDK 1.7.0)
Git or Mercurial
Maven
Jenkins
Now I need to to the following:
At given intervals (eg daily at midnight) is the latest revision in the version control system (tip, HEAD, ...) compiled with Maven. In addition, Java Docs and packages (jar, war) need to be created.
Then Jenkins with all tests conducted and reported.
Make sure there is a report of previous builds
Ensure that the Java Docs and packages can be downloaded (jars, wars, ...) of the latest build
I can't use a GUI on CentOS Minimal so I need to configure the job in xml files? Could please someone show me the way... I'm not a linux server guru.
It's a bit impractical to configure Jenkins via XML by hand, because Jenkins' configuration is spread over multiple files, and the format of the configuration files changes between releases.
Given that Jenkins is a web application, you should be able to visit port 8080 (Jenkins' default port, assuming you didn't change it) on the server where you installed Jenkins (e.g. http://mycentosserver.example.com:8080), and configure it via the web interface.
If you're unable to access the web interface because of a firewall or similar, but you are able to SSH to the server (presumably you can, given that you were able to install stuff on it), you could set up an SSH tunnel to forward a port on your local machine to port 8080 on the server. For example, from your local machine, run the following command. You will then be able to access Jenkins on your local machine at http://localhost:28080 . If you're on Windows, you can use Putty to do the same thing.
ssh -L 28080:127.0.0.1:8080 mycentosserver.example.com
If you can't access the web app directly, and you can't SSH tunnel, I'd recommend setting up Jenkins on a server where you can access the web app, configuring it, and copying the XML config files from /var/lib/jenkins on that server across to your Centos server.