Error executing Workflow "MetaDataModifiedException"

hi ​@REOGIRARD ,

one reason for such a thing to happen is if an API used inside the workflow got updated (metadata changed) and the projection cache is not yet refreshed. 

ref:Considerations Before Designing a Workflow - Technical Documentation For IFS Cloud

If this is the case there is an improvement done on this area that should mitigate this error and its available from 24R2 SU11, 25R1 SU5, 25R2 +

Hi @kamn

Thank you for you answer.
That’s what I was thinking, but we did not updated APIs as it’s IFS standard even if we are on premise hosting.
It might be related to our customs but I don’t remember I made an update, and customs involved in the workflow are mine, for this specific case.
Is the projection cache related to dictionary cache ? From IFS aurena, there is no projection cache refresh option.
We gonna upgrade IFS to next SU then releases once the go-live happens for sure, hope this will help in future.

Thank you, have a nice day.

Guillaume

Hello,

We are still experiencing the issue.

We deployed a new TEST environment for our collaborators, and we are still enhancing it with additional functionalities and fixes.

The error still occurs, even after a full cache refresh (Dictionary, Reference, Security, Object Connection).

Is there another way to refresh the projection cache specifically?

Thank you,

Guillaume

Hello,

We are still experiencing the issue, and it continues to significantly impact our business process.

Previously, we added a client cache refresh button using the following URL:

domain.com/main/ifsapplications/web/server/clearmetadatacache?ALL

At first, this helped resolve the metadata desynchronization issue. However, it is no longer effective, and we are still encountering metadata errors.

For context, our workflow performs the following actions:

• adds operations to Shop Order
• updates Shop Order operation values
• creates new parts (sales, purchase, inventory, and master) through the Copy Part module
• creates product structures
• creates routings
• adds work guidelines to newly created routings
• updates the Shop Order Materials tab
• creates suppliers linked to purchase parts, with pricing calculated on our side

We understand that this is a large workflow, and the workflow tool appears too limited for this use case. Debugging takes around 10 minutes to load, the OData pod crashes very frequently, and the process is highly sensitive to any projection change, which seems to trigger the metadata issue again.

However, when there is no metadata issue, the workflow works perfectly and performs quickly.

At this stage, we cannot upgrade to 24R2 SU11 because we must first complete our migration from IFS V8 to 24R1.

Could you please advise what solution or workaround can be applied to prevent or avoid this metadata issue in our current version?

Best regards,
Guillaume

​@REOGIRARD  could you please share your workflow?

When you mention an HTTP call, are you using the REST connector or an API projection call?

At which stage does the error occur — READ, CREATE, or UPDATE?

Are you running the workflow directly, or in asynchronous mode?

Is the projection standard?

Is it possible to refactor the workflow? and change the trigger?

Kind Regards

Hello ​@N.GEORGI ,

Sadly, I cannot provide the workflow itself, as it is part of our company’s know-how and internal process within a demanding and strategic industry in our country.

The HTTP call used for the client cache refresh is outside the workflow itself. For debugging purposes, I implemented a Command Group for this action. At first, it worked when another user used this button, but when I used it myself, I still received the metadata error.
The workflow BPA itself only performs data processing through JavaScript scripts and internal IFS API calls.

I am unable to determine exactly when the error occurs. In the Trace tab of the DevTools extension, I can only see the BPA called by my own BPA, because we have many customizations across multiple IFS objects. As a result, I cannot identify which projection is causing the issue.

The workflow runs directly and is triggered by a Command Group.

All projections involved are standard projections with customizations. For example, on the Shop Order entity, we added a linked Tools entity to display a tab containing all tools used during the shop order, without requiring users to expand each Shop Order operation, which is the standard behavior in IFS Cloud 24R1.
We also use custom projection actions that call methods from referenced Oracle API packages, so even our standard projections are customized, which is natively supported by IFS Cloud.

It may be possible to refactor the workflow, as it is sequenced, but the workflow triggered by the Command Group calls several sub-BPAs to complete the overall process.

I do not know whether splitting the process further into additional sub-BPAs would improve the situation.

It is also difficult to test each step independently. For example, routing creation requires the parts to have already been created, and product structure creation depends on the routing being completed first. These dependencies were inherent to the workflow design.

Best regards,
Guillaume

If a BPA refactor isn’t feasible, perhaps consider refactoring the functional process instead ? (if possible based on your working needs)

If several sub BPA are already present, this could be an ehacment to adress.

It’s difficult to provide more specific guidance without knowing the workflow.

In 24R2, OData metadata handling has been improved, so your issue may already be resolved.