Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
demo: wait for stopper to stop on shutdown
Previously, attempting to quickly shutdown and restart a node using the demo `\demo shutdown` and `\demo restart` command would result in an error: ERROR: internal server error: failed to create engines: resource temporarily unavailable This was revealed in the multi-tenant tests but is not multi-tenant specific and happens in non-mt demo clusters as well. The `\demo shutdown` command sends a Drain request with `Shutdown: true`. However, as part of the shutdown, the gRPC server is shut down, which will return an error to the client (which is ignored) before the entire server is shutdown. There is a TODO in the code from 2019 asking why we don't shut down gRPC later in the process. This is a demo cluster specific fix in which we wait on the stopper client side since we happen to have a reference to it. This is intended for backport so the disabled test can be re-enabled on the release branches. It's likely that we can do a larger fix in which we leave the drain server running for as long as possible. But, the blast radius of such a change would not be suitable for backport given the low-impact of the problem. Fixes: #110748 Release note: None
- Loading branch information