Hello! Is this normal to get a jump in the autoinc...

# exposed

[Stephen Vinouze]

Hello! Is this normal to get a jump in the autoincrement value while batch upserting? For instance, I batch insert the first time about 30k of rows. On the second request, I do the same but only 10k of rows are added this time (others are duplicates). But these new inserted are created with a id of starting at about 90k. I suppose it’s due to the batch that allocs an autogenerated id even though no row was inserted but wanted to confirm with you. Especially because it means I may need to reconsider the way I handle ids or I may reach my bigint faster than I anticipated 🙂

[Damien O'Hara]

Not strictly Exposed related but this sounds like typical behavior for auto_increment. there are many circumstances where auto_increment ids will be allocated including when an upsert batch is being prepared or when a transaction with auto_incremented ids is rolled back. same goes with Postgres sequences/serial

[Stephen Vinouze]

Thanks for your input. That confirms what I was thinking and makes somehow sense to me as well. I’ll handle my id with UUID then 🙂