Step 1

In this tutorial, we continue where we left off in integrating our first job type, Hello World. We will update our job type to generate a HySDS dataset that will be automatically ingested and catalogued into the system.

1. On the mozart instance, navigate to the hello_world repository

   cd ~/mozart/ops/hello_world

2. We need to author a dataset ID scheme according to the HySDS dataset specification. In this tutorial, we will use the following dataset ID scheme

   hello_world-product-<year><month><day>T<hour><minute><second>Z-<hash>

   where <year> will be replaced by a fictional observation year and likewise for <month>, <day>, <hour>, <minute>, <second> and <hash>. An example instance of this dataset ID would be

   hello_world-product-20170901T000102.021Z-a3d0x

3. Edit the script that runs our job type

   and add code to create an instance of our dataset. The script should look like this

   N.B. In production, the hash should be generated from the input parameters (or simply the contents of _context.json), for deduplication purposes. The required fields are:

   • version

   The optional fields are:

   • label

   • location (in GeoJSON "geometry" format)

   • starttime

   • endtime

4. Access the documentation on the specifications for the HySDS dataset JSON file and metadata JSON file for more detail. Finally, commit your change and push the update

   If your jenkins job for the master branch of your hello_world repo was configured correctly, the git push should have kicked off a rebuild of the container. If not, navigate to your ciinstance (e.g. http://:8080) and manually schedule a build

5. Test the script by running it

   You should see the dataset directory you just created sitting there in your directory. Ensure that the fake_data.dat and the HySDS dataset and metadata JSON files were generated

6. Now let's try to ingest this new dataset into our catalog

   You should've gotten an error that the data was not recognized

7. Let's configure our new dataset so that it can be recognized by our HySDS cluster. Since we want to ensure all nodes in our cluster is able to recognize this new dataset, we need to make updates to the datasets.json file in ~/.sds/files. Still on mozart

   edit the datasets.json file

   and add the following dataset configuration below the dumby-product configuration

   and save.

8. Run the following fabric command to push the update you made to the datasets.jsontemplate on mozart

   You should have output similar to the following

9. Now let's try to ingest this new dataset into our catalog

10. Verify that your new dataset was published to your HySDS cluster. Open up a browser and point it to your grq instance's tosca (Dataset FacetSearch) interface (e.g. https:///search). On the left panel you should see your new dataset type listed under the "dataset" facet showing 1 result

    Click on "hello_world-product" under the "dataset" facet and your result set should be constrained now to just the dataset you just ingested

11. Now that we've verified that the new dataset is recognized by the system, let's push the updates to the rest of the system. On mozart

That's it!

Congratulations, you've integrated a new dataset type into your HySDS cluster.

To configure Autoscaling groups for your HySDS cluster, continue on to Create-AWS-Autoscaling-Group-for-Verdi.

To configure a staging area for your HySDS cluster, continue on to Create-AWS-Resources-for-Staging-Area.