Roy Dahan

Optimize test setup time

10

Test setup include several steps like provisioning instances, setting up db nodes, loaders, monitor. Some steps are doing installations that may not be necessary or could be replaced with an...

Change CI version fetcher to pull only released versions as the RC releases may be unstable to be tested with. These releases should be tested as part of driver-matrix tests...

Add support for user schema

8

It will be really helpful if we can add support to use user schema like we have in c-s.

Consider using capacity reservation

2

In order to overcome "Insufficient Capacity" in several places, espcially where we try to use placement groups, we can try using "capacity reservation" - https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/capacity-reservations-using.html#capacity-reservations-create

perf_simple_query - tps and tasks_per_op regression

4

perf_simple_query results from begining of may shows regression in tps and tasks_per_op: Test results | commit_id | date | run_date_time | version | allocs_per_op | instructions_per_op | mad tps |...

gocql implementation of "Decouple schema fetch queries from server-side timeouts"

1

see https://github.com/scylladb/scylladb/issues/18557

Python implementation of "Decouple schema fetch queries from server-side timeouts"

1

see https://github.com/scylladb/scylla-drivers/issues/4

As supported in all other drivers (e.g. https://github.com/scylladb/scylla-rust-driver/issues/604 and https://github.com/scylladb/java-driver/pull/200) we should have it supported also in Python driver.

https://github.com/scylladb/scylla-bench/blob/e87621070b86ad0c8e1d7ae78500414c55895f14/main.go#L137-L146 In case the user supplies: "-partition-count=50" and "-concurrency=100" it seems that the first 99 threads won't do anything and the last thread will read all partitions.

Add a performance test for Rust Driver

1

We need (if we don't have) a performance test to measure performance for driver and regularly run it to detect performance regressions.