0
Under review

Joins to non-existant adapter objects created when outgoing baseline syncs fail to provision in the connector

Adrian Corston 8 months ago in UNIFYBroker/Plus updated 3 months ago 6

I have a customer where connector entity create fails (due to external issue on the customer side) for a number of entities during link sync, and for each one I end up with a Join showing on the Locker but no corresponding adapter entity.

For example, for a baseline sync I see:  Update entities 1690 to connector AD User reported 1690 entities saved, 3 failed

After this I look at locker joins:  Showing 1 to 10 of 1,693 entries

But adapter joins says: Showing 1 to 10 of 1,690 entries

This causes problems on the next baseline sync, because even though the locker is already joined (to a non-existant adapter entity) another attempt to create the adapter/connect is still made but then this error appears:

Unify.Product.Plus.JoinExecutorTargetEntityAlreadyJoinedException: Source entity '2ac4395a-2cb7-4f6e-9d80-e6a6a38994b7' value matched to target entity 'f70abdcf-4097-4d41-84a6-b0703308f16f', however this target entity is already connected to another source entity. 'fb9d3fea-9b4a-4708-a0e1-e730551ef030'. Cannot proceed with join.

I thought we worked through this and already had a patch for this issue, but I cannot work out which ticket it was in.

Could you please investigate and see what the situation around this is?

Under review

Hi Adrian,

Do the phantom locker joins disappear at all after any actions, such as a confirming connector import and changes sync? Does removing the joins manually just repeat the same pattern?

Import All on the connector does not clear the phantom locker joins.

After manually deleting a phantom join and re-running the baseline sync the phantom join reappears.

Note: this issue is not urgent priority since it's only happening for users who can't provision to AD due to data misconfiguration by the customer.

Perfect, thanks. I can find similar tickets which have patches already in the environment, but not one which describes this issue directly. I'll take a look at see what I can find 

The data has been remediated (full reload) since the P1 needed to be resolved and the customer's service restored.  I should still be able to simulate the issue in a non-Prod environment if it comes to testing a fix though.