Open keith-turner opened 5 years ago
This seems like a good idea. Some of the internal code that is being used is being removed. For example, see this comment.
if anyone is interested I can provide pointers within the Fluo code.
@keith-turner that would be great in case I get around to working on this at some point.
@mikewalch, I see you are assigned to this. Are you still looking into this or did you make any progress here?
No, I am not looking into this. Feel free to work on it.
@DomGarguilo Fluo is multi-module project like Accumulo, its config may be a bit more complicated than you need. So the base pom defines the plugin below
https://github.com/apache/fluo/blob/44e134a9d90685b57fc60a8bcc9a8919e79e4e3b/pom.xml#L294-L333
Then in a module plugin it actives it as follows. It does this in a profile so that when test are not running it does not start miniaccumulo.
Its important to define the accumulo,hadoop, zk, and thrift versions outside the plugin. The maven plugin is built against Accumulo 2.0 and if this is not done it may use older version in the launched processes.
The following code gets creates an accumulo client in the test. It uses the client properties file created by mini accumulo that the plugin started.
The maven plugin is built against Accumulo 2.0
I am working towards getting accumulo-proxy to work with 2.1 so if the plugin does not yet support 2.1 then maybe getting that working should be the next step
Edit: I see that the plugin does allow for version 2.1
Edit: I see that the plugin does allow for version 2.1
Working with the accumulo-maven-plugin has taught me a lot about Maven. Each plugin has its own classpath independent of the projects classpath. So, the accumulo-maven plugin will do the following.
Currently the proxy uses internal Accumulo code for ITs. It could use the Accumulo maven plugin for running ITs instead. Fluo uses this plugin for its ITs, if anyone is interested I can provide pointers within the Fluo code.