Management API Error: delete-objects-batch failed

[OSMOZ]

Hello,

I am unable to delete some objects via the delete-objects-batch call, I get the following error:

Batch operation failed. java.lang.RuntimeException: an eclipse error has occurred enable logging on EclipseLinkExceptionHandler to see full error; nested exception is javax.persistence.PersistenceException: java.lang.RuntimeException: an eclipse error has occurred enable logging on EclipseLinkExceptionHandler to see full error

 The request payload is just a list of group names:

{"objects":[{"type":"group","list":[{"name":"my_first_group"},{"name":"my_second_group"}]}]}

This call is made on the active server in a MDS domain running version R81.20. (Check Point Security Management Server R81.20 - Build 440)

Looking online I found an SK: https://support.checkpoint.com/results/sk/sk152592
However there are no pending changes and restarting the mds processes does not seem to solve the issue.

I cannot see any errors in the $FWDIR logs.

EDIT: The issue only occurs when there are 5 or more objects to be deleted, 1 to 4 objects works fine. The issues does not seem to be linked with a specific object.

EDIT 2: When reaching more than 100 objects to be deleted in the payload the error is masked and the task shows a "Batch operation completed successfully" message even though the operation failed (objects were not deleted).

Does someone have any leads where to search for ?

Thanks

[_Val_]

Single command or a script? Can you delete an object with the same command, if only a single group is in the list?

[OSMOZ]

This request was sent both manually and via a script but I got the same result on both.

I did manage to delete a single object via the batch api.

Would this indicate that one of the groups is causing the issue ? And if so would there be a way to see which one in the server logs ?

[OSMOZ]

Experimenting further I figured that the issue only occured when there were 5 or more objects to be deleted, 1 to 4 objects works fine. The issues does not seem to be linked with a specific object.

[OSMOZ]

Continuing my tests I found a new edge case :
When reaching more than 100 objects to be deleted in the payload the error is masked and the task shows a "Batch operation completed successfully" message even though the operation failed (objects were not deleted).
Hoping this could help narrow down the cause.

[PhoneBoy]

I suspect the issue is that you're trying to do too many things at once without a "publish" operation.
We generally recommend this after 100 operations or so.
I would think the batch APIs would handle more than this, which is why it's probably worth a TAC case to investigate: https://help.checkpoint.com

However, I suspect you'll have to do this in smaller batches. 

[PhoneBoy]

Anything you're aware of @Omer_Kleinstern?