Stan Rosenberg

Smoke test results look good! Resolving merge conflict and merging. Thanks for making roachprod ready for external use!

Final smoke test [SELECT_PROBABILITY=0.25](https://teamcity.cockroachdb.com/buildConfiguration/Cockroach_Nightlies_RoachtestNightlyGceBazel/15462087?buildTab=log&logFilter=debug&focusLine=6&logView=linear&linesState=3), since there were non-trivial merge conflicts after [1] got merged. [1] https://github.com/cockroachdb/cockroach/pull/124604

Final smoke test looks good, let's 🚢 bors r+

> However, that turned out to hit a wall: we already have `option.VirtualClusterName` defined in the roachtest `option` package, but that function is defined as an option for `c.Conn{,E}`. Could...

> We could do something like this. I considered that option, but the reason I didn't like it as much is that then you give up on any kind of...

> I like this solution more, but would this fall apart if we ever need to use these options in a non connection context? i.e. if we add function that...

> The option definitions are a lot simpler and without the `applyFunc` business which is a big win. I concur, definitely a big win; the user-side is a lot more...

> Before this patch, we'd segfault if cluster creation fails and we're running with --debug-always. > > This patch is a small protection against the problem, but I don't think...

Since this has come up in the context of performance benchmarks, we should be careful not to retry excessively. In some sense, a benchmark becomes tainted since retries could yield...

Failure as of the run in [1], ``` | | ./cockroach workload init tpcc --regions=us-east1,us-west1,europe-west2,europe-central2 --partitions=4 {pgurl:1} --db tpcc_background --warehouses 600 | | ``` | | ... took 11h29m6.622186325s, 0.00...