Israel Fruchter

> > I guess we can easily cache this property, it's not something that can change during a test, we never reconfigure networks in the middle of a test >...

> Due to public IP address changes after nemesis-initiated node restarts, caching mechanisms become ineffective, and reconfiguration is required. This is the reason why I did not cache it but...

the property `network_configuration` that @vponomaryov is pointing to in this issue. that can be cached, isn't it ?

seems like the mapping that was introduced in: https://github.com/scylladb/scylla-cluster-tests/pull/5869/ might contribute to some of the slowdown, since it calls `get_all_ip_addresses()` for each not every time it created we might need...

let's wait and see how it would be the fix (it's a one liner, I hope we'll get it fast enough) not sure it would worth our while to refactor...

> > @roydahan, @vponomaryov - @dkropachev is already working on fixing the problem with the nemesis. You can monitor the progress in #7029. This PR should be closed. > >...

at least for one of those `test_single_interface`, I can say it can't even build a cluster ``` # Since scylla CCM doesn't yet support this options, we skip it #...

@aleksbykov you added this inside ` if self.validate_large_collections:` clause so it's not exactly globally as the title says, it would work only if a case has validation for large collection...

@timtimb0t which version of s-b ? if `0.2.0`, please re-run it

> @fruch , it is 0.2.0 but the test was for rc4 and investigation should have been finished today, so there is no sense to rerun I think it is...