Hi Team,
We have 8GSCs deployed with 20 partitions n 20 backups. Now While events were being consumed our GSC crashed bcoz of PermGen issue. while 3 events were in taken from space under transaction. Ideally backup should have these events until transaction is not completed and it should become primary and these events should be again processed. However we lost these event. Note here is that the our upstream publishes data in a non-transactional manner.
Can someone please provide their insights why we had lost the data and what can be done to mitigate the risks as we cannot go in production with this risk and will become a blocker for us.
↧