Esti system test to verify a system wide migration to KV, for correctness
Can also be used to cover the migration from the performance aspect by (currently manually) checking the time it takes
As the migration process will increase over time, as we transfer more parts to KV, the time to migrate cannot be well defined
We will use this to identify "unreasonable" migration time
The main idea is to have a system running on SQL DB, performing various lakefs actions [TBD] and bringing lakefs to a certain state. Then lakefs is stopped, a migration to KV takes place and lakefs is restarted with KV enabled. The testing flow should continue at this point, seamlessly
Note: Currently, Esti tests rely on lakefs being up and running prior to the test itself. This task needs to add support for stopping/starting lakefs as part of the test
The time of the migration process should be "benchmarked" as part of this test. Initially, this can be done by manually verifying the migration time
What is the sufficient dataset prior to migration?
Esti system test to verify a system wide migration to KV, for correctness
The main idea is to have a system running on SQL DB, performing various lakefs actions [TBD] and bringing lakefs to a certain state. Then lakefs is stopped, a migration to KV takes place and lakefs is restarted with KV enabled. The testing flow should continue at this point, seamlessly
Note: Currently, Esti tests rely on lakefs being up and running prior to the test itself. This task needs to add support for stopping/starting lakefs as part of the test
The time of the migration process should be "benchmarked" as part of this test. Initially, this can be done by manually verifying the migration time
What is the sufficient dataset prior to migration?