--- title: "32-bit integer primary keys cause processing delays" slug: "32-bit-integer-primary-keys-cause-processing-delays" updated: 2025-10-20T20:02:55Z published: 2025-10-20T20:02:55Z canonical: "help.hyperscience.ai/32-bit-integer-primary-keys-cause-processing-delays" --- > ## Documentation Index > Fetch the complete documentation index at: https://help.hyperscience.ai/llms.txt > Use this file to discover all available pages before exploring further. # 32-bit integer primary keys cause processing delays ## Issue Historically, application database tables were created with 32-bit integers as primary keys. This approach has proved insufficient for large volumes of submissions and may cause the application to become unresponsive. This issue was addressed in v33.1 and v34, so any new installations have 64-bit keys by default. However, the issue still exists in previous versions of Hyperscience. ## Affected versions Versions prior to v33.1 Also, if your instance’s original version is earlier than v33.1, and you upgraded to a more recent release, your instance is still affected. ## Affected configurations All configurations ## Fixed versions - v33.1 and subsequent versions of v33 - v34 and later ## Cause Hyperscience used 32-bit integers as primary keys in older versions. ## Solution / Workaround We’ve implemented a new utility called `migrate_to_bigint` to migrate primary keys from 32-bit to 64-bit integers. ### 1. Check whether migration is necessary. Run the following command: ```plaintext run.sh manage migrate_to_bigint --dry-run ``` The command outputs an SQL migration plan. If the plan is empty, your database is already migrated, and no further steps are required. An example of an empty plan appears below. ```plaintext -- Stage: DROP_INTERTABLE_CONSTRAINTS -- Stage: ALTER_TABLES -- Stage: ADD_INTERTABLE_CONSTRAINTS ``` ### 2. Run the migration using run.sh. 1. Stop all Hyperscience instances sharing the same database. 2. Run `run.sh manage migrate_to_bigint`. Note that, depending on database size, the migration may take multiple hours to complete. We encourage you to plan appropriate maintenance windows ahead of time. We recommend that no other resource-intensive tasks, including backups, occur at the same time as the migration. To ensure faster operations, provision additional IOPS, and allocate an adequate disk-space buffer.