Upgrade 12.6 -> 13.4: Empty nodes bootstrapped into repo

Hi there

When I deploy our updated BrXM project to one of our AWS environments, most content and configuration is replaced by empty nodes or stuff coming from the implementaion project.
I assume these empty nodes are the ones coming from the product (core/hippo-cms/hippo-repository/hippo-repository-workflow). I am expecting that only relevant changes are bootstrapped and existing content are left in place.

Some examples on what happens after deploy

  • documents, gallery, asset nodes are completely empty
  • hst:host only contains whats in the implementation project. Other hosts are configured before deploy.
  • hst:users only contain whats in the implementation project - lots of users before deploy
  • After deploy there is a “upgrade-to-12-content-marker” node in the console, which seems off to me since we’re upgrading to 13

I’m guessing something is off with the config baseline, but I have no clue what to try by now.

Only error I see in the logs are:
ERROR org.onehippo.cm.engine.ConfigurationContentService - Processing ‘APPEND’ action for content node ‘/content/documents/dk-sites/administration’ failed.
javax.jcr.PathNotFoundException: /content/documents/dk-sites

Locally the updated project is running fine. And we deploy authoring and delivery separately to AWS.

Tips and trick for stuff to try are most welcome :slight_smile:

Hi Ulrich,
This lies somewhere in the bootstrapping phase on start-up. First thing to check is whether the delivery nodes have system property repo.bootstrap=false startup option. Bootstrapping should be the responsibility of the CMS.

Bootstrapping can be followed in detail by setting log levels to INFO, DEBUG of classes:

  • org.onehippo.cm.engine.ConfigurationContentService
  • org.onehippo.cm.engine.ConfigurationServiceImpl

I don’t think ‘upgrade-to-12-content-marker’ node is something from the product, guessing this is project code base related.

HTH
Jeroen

Just a very late update (Sorry). You lead us in the right direction. Tried setting the log level, but it had no effect. We discovered that all the catalina options we set in our build scripts was not exported for the instance to see… doh! Solved a lot of issues for us, thanks!

1 Like