Problems with the retraction wizard

[Nicklas Nordborg]

The initial symptom is that the "Mark extracts for retraction" wizard hangs after entering a SCAN-B id. There is no error message or progress information. It should typically be very quick to bring up the next step in the wizard. After about an hour the next step finally shows up and the list with SCAN-B id seems to contain more or less all registered cases that we have.

Investigations indicate that there is a problem with the recursive loading of "all child items" in RetractionServlet.collectAllChildItems() method. It is working and seems to be collecting several hundred thousand items as child items.

The theory is that the problem points are steps in our pipelines where items are merged and analyzed together. The current implementation have special case for PooledLibrary items. This was probably the only case when the retraction functionality was implemented, but since then we have added more pipelines that merge items. For example, BeadChip items used for the genotyping and methylation pipelines. There seems to a problem also with the ArrayDesign link since we moved all raw bioassays to derived bioassays.

[Nicklas Nordborg]

In 8087:

References #1671: Problems with the retraction wizard

Implemented a check for item type so that some item types can be ignored ( ARRAYDESIGN, PROTOCOL, SOFTWARE and JOB).

BeadChip items are added as child item but further processing is not done. They need to be handled similar to pooled libraries.

[Nicklas Nordborg]

In 8088:

Fixes #1671: Problems with the retraction wizard

This should fix the current problems with the retraction wizard.

In the future we need to be careful when adding new pipelines that may introduce other item types where items are handled together since they typically need special handling.