mirrors/scylladb
1
0
Fork 0
mirror of https://github.com/scylladb/scylladb.git synced 2026-05-31 03:56:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

storage_service: topology coordinator: rollback on handle_tablet_migration failure during tablet_draining stage

Browse Source 
During remove or decommission as a first step tables are drained from
the leaving node. Theoretically this step may fail. Rollback the
topology operation if it happen. Since some tables may stay in migration
state the topology needs to go to the tablet_migration state. Lets do it
always since it should be save to do it even if there is no on going
tablet migrations.
This commit is contained in:
Gleb Natapov
2023-11-21 14:16:40 +02:00
parent 7267376eac
commit 4ebdddc31b
1 changed files with 11 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
service/storage_service.cc
View File

@@ -1979,7 +1979,16 @@ class topology_coordinator {
}
break;
case topology::transition_state::tablet_draining:
co_await handle_tablet_migration(std::move(guard), true);
try {
co_await handle_tablet_migration(std::move(guard), true);
} catch (term_changed_error&) {
throw;
} catch (group0_concurrent_modification&) {
throw;
} catch (...) {
slogger.error("raft topology: tablets draining failed with {}. Aborting the topology operation", std::current_exception());
_rollback = true;
}
break;
case topology::transition_state::write_both_read_old: {
auto node = get_node_to_work_on(std::move(guard));
@@ -2346,6 +2355,7 @@ class topology_coordinator {
topology_mutation_builder builder(node.guard.write_timestamp());
builder.set_fence_version(_topo_sm._topology.version) // fence requests in case the drain above failed
.set_transition_state(topology::transition_state::tablet_migration) // in case tablet drain failed we need to complete tablet transitions
.with_node(node.id)
.set("node_state", node_state::normal);