storage_service: topology coordinator: log rollback event before changing node's state

The test for the rollback relies on the log to be there after operation
fails, but if node's state is changed before the log the operation may
fail before the log is printed.

Fixes scylladb/scylladb#15980

Message-ID: <ZUuwoq65SJcS+yTH@scylladb.com>
This commit is contained in:
Gleb Natapov
2023-11-08 18:00:34 +02:00
committed by Kamil Braun
parent d8b6771eb8
commit 2dd8152c8b

View File

@@ -2467,8 +2467,8 @@ future<> topology_coordinator::rollback_current_topology_op(group0_guard&& guard
auto str = fmt::format("rollback {} after {} failure to state {}", node.id, node.rs->state, state);
co_await update_topology_state(std::move(node.guard), {builder.build()}, str);
slogger.info("{}", str);
co_await update_topology_state(std::move(node.guard), {builder.build()}, str);
// Try to run metadata barrier to wait for all double writes to complete
// but ignore failures
try {