Israel Fruchter

closing until someone would need to attend to it again

> do you have some proof of this helping out? If not, I/You could take it for a ride in elasticity-test where I saw lower throughput than c-s ([scylladb/cql-stress#98](https://github.com/scylladb/cql-stress/issues/98)) not...

> > testing here: https://jenkins.scylladb.com/view/staging/job/scylla-staging/job/lukasz/job/elasticity-test/8/ > > Works, but didn't change anything (I looked at preload stage - took same time) and throughput/latency was the same ?

> > > testing here: https://jenkins.scylladb.com/view/staging/job/scylla-staging/job/lukasz/job/elasticity-test/8/ > > > > > > Works, but didn't change anything (I looked at preload stage - took same time) > > and throughput/latency...

> > > > > testing here: https://jenkins.scylladb.com/view/staging/job/scylla-staging/job/lukasz/job/elasticity-test/8/ > > > > > > > > > > > > Works, but didn't change anything (I looked at preload stage...

base on results we seen in: https://github.com/scylladb/qa-tasks/issues/1715#issuecomment-2290683967 seems like this is actually helping out, and level the docker loaders with VM loaders.

> We could also try with not providing public ip at all in Azure That specific per instance > > also I'd rather see a field in `InstanceDefinition` instead changing...

> so could apply additional rule to block all with higher priority that's not a new rule, that's a new NSG, that shouldn't be applied to runner

@soyacz already implemented this one closing

> We need to think what to do in the case when we terminate a node and expect it to be almost immediate operation. If it adds a significant amount...