The new version seems to be working like it should now.
Before running there were 5909 action.hold_copy_map entries associated with frozen holds, and after running with --retarget-frozen there are 10395 entries.
It did result in 5 out of the 19 frozen holds that had expiration dates set to be canceled. I tried suspending a few holds in the 3.10 angular and angularjs interfaces and in all cases the expire_time was set to null when suspending/freezing the hold. So the few holds in our production data must be from an old interface or some manual action. I asked on IRC and Jeff Davis said he also had a number of them in their 3.9.1 system, they were all several years old though, so nothing recent. So probably whatever bug caused them was from a while back.
I'll work on adding a DB upgrade script to fix any instances left over to prevent any frozen holds from being canceled unexpectedly.
The new version seems to be working like it should now.
Before running there were 5909 action. hold_copy_ map entries associated with frozen holds, and after running with --retarget-frozen there are 10395 entries.
It did result in 5 out of the 19 frozen holds that had expiration dates set to be canceled. I tried suspending a few holds in the 3.10 angular and angularjs interfaces and in all cases the expire_time was set to null when suspending/freezing the hold. So the few holds in our production data must be from an old interface or some manual action. I asked on IRC and Jeff Davis said he also had a number of them in their 3.9.1 system, they were all several years old though, so nothing recent. So probably whatever bug caused them was from a while back.
I'll work on adding a DB upgrade script to fix any instances left over to prevent any frozen holds from being canceled unexpectedly.
Josh