If a start a batch to migrate 4.000.000 process instances currently the batch seed job will try to create all 4.000.000 batch jobs even if they aren't processed yet. This might lead to a huge jobs table. Some operations fetch all jobs for a batch or count them. This might lead to performance issues.

AT:

• it exists a process engine configuration which limits the number of batch jobs which limits the number of jobs for a batch at any time
• the seed job never creates more jobs than this configuration
• (maybe) the seed job should than use a back off strategy to not to often poll the jobs count