There seems to be some overlapping context between this report and bug #1524030. If we reduce the number of revocation events to only the events we absolutely need there will be less bloat in the revocation event table overall.
How long does it take your deployment to generate 30,000 revocation events (days, weeks, months)?
I believe you specifically referencing this code:
https:/ /github. com/openstack/ keystone/ blob/1254ac0283 32778d086a7884a ab1fe7c60748590 /keystone/ revoke/ backends/ sql.py# L116-L125
There seems to be some overlapping context between this report and bug #1524030. If we reduce the number of revocation events to only the events we absolutely need there will be less bloat in the revocation event table overall.
How long does it take your deployment to generate 30,000 revocation events (days, weeks, months)?