c/balancer_backend: first initialize planner and then call plan

This change is a part of an effort to identify and fix rare segmentation fault in Redpanda that happens after it was suspended with `SIGSTOP` signal. According to the C++ standard the temporary should be kept alive until the expression ends. The crash we are observing indicates the UAF issue. The only way the variable, that access causes the segfault, can be deleted is by getting out of scope which in this situation should be guaranteed. Given our experience with coroutines and different types of lifecycle bugs that we found in past this is a poor man's effort to avoid the issue. Signed-off-by: Michał Maślanka <[email protected]>
redpanda-data · Apr 26, 2024 · a52d0ad · a52d0ad
1 parent 4182b85
commit a52d0ad
Showing 1 changed file with 20 additions and 18 deletions.
diff --git a/src/v/cluster/partition_balancer_backend.cc b/src/v/cluster/partition_balancer_backend.cc
@@ -357,24 +357,26 @@ ss::future<> partition_balancer_backend::do_tick() {
     // claim node unresponsive it doesn't responded to at least 7
     // status requests by default 700ms
     auto const node_responsiveness_timeout = _node_status_interval() * 7;
-    auto plan_data
-      = co_await partition_balancer_planner(
-          planner_config{
-            .mode = _mode(),
-            .soft_max_disk_usage_ratio = soft_max_disk_usage_ratio,
-            .hard_max_disk_usage_ratio = hard_max_disk_usage_ratio,
-            .max_concurrent_actions = _max_concurrent_actions(),
-            .node_availability_timeout_sec = _availability_timeout(),
-            .ondemand_rebalance_requested
-            = _cur_term->_ondemand_rebalance_requested,
-            .segment_fallocation_step = _segment_fallocation_step(),
-            .min_partition_size_threshold = get_min_partition_size_threshold(),
-            .node_responsiveness_timeout = node_responsiveness_timeout,
-            .topic_aware = _topic_aware(),
-          },
-          _state,
-          _partition_allocator)
-          .plan_actions(health_report.value(), _tick_in_progress.value());
+
+    partition_balancer_planner planner(
+      planner_config{
+        .mode = _mode(),
+        .soft_max_disk_usage_ratio = soft_max_disk_usage_ratio,
+        .hard_max_disk_usage_ratio = hard_max_disk_usage_ratio,
+        .max_concurrent_actions = _max_concurrent_actions(),
+        .node_availability_timeout_sec = _availability_timeout(),
+        .ondemand_rebalance_requested
+        = _cur_term->_ondemand_rebalance_requested,
+        .segment_fallocation_step = _segment_fallocation_step(),
+        .min_partition_size_threshold = get_min_partition_size_threshold(),
+        .node_responsiveness_timeout = node_responsiveness_timeout,
+        .topic_aware = _topic_aware(),
+      },
+      _state,
+      _partition_allocator);
+
+    auto plan_data = co_await planner.plan_actions(
+      health_report.value(), _tick_in_progress.value());
 
     _cur_term->last_tick_time = clock_t::now();
     _cur_term->last_violations = std::move(plan_data.violations);