While updating to the latest versions of clair (v4.3.6) and claircore (v1.2.0), claircore had a new libindex migration step (
https://github.com/quay/claircore/blob/08db7b99956c50c8ccd01d6039df9c1ad86eae48/libindex/migrations/04-foreign-key-cascades.sql) which we had real problems getting to run. In some of our larger instances it took longer than 1 hour, which hit some enforced timeout we have (doesn’t seem something we have direct control of as we tried to increase our timeouts), which means the transaction did not complete. We got around this by splitting up the migration into smaller steps, running them directly and then updating the libindex migration table.
I wanted to ask that if possible could migrations be done in smaller steps going forward to avoid these situations? Thank you!
Paul