I think that some latest code in Master branch has introduced a regression in GeoServer Layer publishing.
I'm wondering if there is some issues with the GeoServer Catalog management (see the final note in this report).
Steps to replicate the issues:
- From the GeoServer GUI, click on Data->Stores->add a new Store
- Select GeoTIFF (I have noticed same problem with ArcGrid so I think it's format unrelated)
- specify a name for the store and browse the disk to refer to a standard GeoTIFF. (You may consider using geoserver sample data GeoTIFF. It's available in the GeoServer source code: geoserver\data\release\coverages\img_sample\Pk50095.tif)
- click on save.
You will notice the first issue:
Usually, you would have immediately seen the list of available layers to be published from the newly configured store.
Right now, you need to manually select the newly configured CoverageStore from the dropdown, instead.
If you configure an arcGrid coverageStore using a TIFF file, the store creation won't complain against that. I think there is some validation issue.
Edit: investigating a bit more on this I have found that:
The clone method doesn't clone the store ID (there is no ID setter method) so that the expandedStore has no ID.
The NewLayerPage(expandedStore.getId()) call will send a null ID so no checks are performed on the store. This is why it doesn't complain about the wrong format and why it doesn't automatically list the available layers.
- select the Previously configured GeoTIFF store from the dropdown
- click on the Publish link for that layer.
- hit save (this should publish the layer)
- select again the "add a new layer" link
- select again the same exact previously configured GeoTIFF store from the dropdown
You will notice that the available layer doesn't have the "Published" checkmark icon.
You may configure other GeoTIFFs and you will see the same result.
If you go to the "ServerStatus" page and click on "Reload" button to Reload the Configuration and the Catalog, all the layers become "Published".