FormDataCleanupJob missing nodes

Hi There,

At a client which im currently working for the FormDataCleanupJob is filling the logs with repository exceptions (ItemNotFoundExceptions). This is caused by strange behavior when running the query from that class:

SELECT * FROM hst:formdata ORDER BY hst:creationtime ASC

The number of results in the repositoryservlet says 525 nodes, but only 25 are shown.

My thought was to run the repository-checker tool to check if these were orphaned nodes. But the result of the repository-checker only resulted in 1 orphaned node. Even when running the tool with some the uuids from the logs. The repository-checker also could not find the nodes.

Why are these nodes found through this query and nowhere else?

With kind regards,
Joost van Zanden

Could it be related to the Lucene index? Are you running a cluster? If yes, can you try to execute the jcr query in different nodes and check the results?

We are running a cluster, and all nodes have the same issues but with difference severities. 1 node has 200 missing jcrnodes, the other 400, and another 500

Try to recreate the index in one of the nodes if possible and then execute the query there.
You could also try some trick like rename/move the root node of the formdata and put it back again to original location/name. That could trigger lucene to re-index the formdata subtree.

depending on number of nodes stored…this could be memory intensive operation…that could lead to OOM error…

Took a little time before i got clearence to try this, but recreating the index on all three nodes in the cluster did the charm. Thanks for the help guys!