AEM workflow instances and archive not always displayed

[sean_gubler]

One of our Adobe tech partners has asked the following....

I’m currently developing an Osgi module for AEM. We are currently using AEM 6.3.3.2 and I’m experiencing an unexpected behavior while using workflows.

Performed steps

1. Create a simple workflow model with only one process step. The process step was implemented by me and packaged in the Osgi bundle

2. Launch the workflow

Expected behavior

The workflow should be displayed in Workflow instances while running and in Workflow archive after a successful completion

Current behavior

I can see from logs that my custom workflow process step ran correctly and that the workflow is displayed in Workflow instances while running, but once completed it is neither in Workflow archive nor in Workflow failures. Moreover this misbehavior is not systematic because sometimes it happened that it was not even displayed in Workflow instances.

Here I’ve captured the node corresponding to the completed instance on CRXDE, which correctly shows the workflow as completed

Additional information

In a local instance of AEM 6.4, this issue is not happening, even if I found on the web some users complaining about a similar issue in AEM 6.4.2 https://forums.adobe.com/thread/2497482

In the logs I’ve found a few exceptions like the ones in the attached file. I don’t know if there’s a correlation.

Can you please support and confirm that this is an issue?

[Andrew_Khoury]

I searched our bug tracker and I couldn't find a matching issue.  However, I have seen some cases where somebody tries to terminate a workflow during an AEM restart and it picks up the Sling POST servlet instead of the workflow servlet.  In these cases this node gets corrupted: /var/workflow/instances

To see if this is the cause and fix it, do the following:

1. Go to http://aemhost:port/crx/de/index.jsp#/var/workflow/instances and log in as admin

2. Check the Properties tab for the properties of the node.  Make sure these properties are set, if they are not then set them:

sling:resourceType=cq/workflow/components/instances

jcr:title=Instances

[arunpatidar]

you can also check -

is workflow transient?

Is workflow purging running frequently?