Error-checking for a couple of corruption issues (#8059)

* Extra error handling
* Handle connection closed in Redis monitor
Co-authored-by: Mehrdad <noreply@github.com>

src/ray/core_worker/core_worker.cc

@@ -152,6 +152,11 @@ void CoreWorkerProcess::EnsureInitialized() {
 CoreWorker &CoreWorkerProcess::GetCoreWorker() {
   EnsureInitialized();
   if (instance_->options_.num_workers == 1) {
+    // TODO(mehrdadn): Remove this when the bug is resolved.
+    // Somewhat consistently reproducible via
+    // python/ray/tests/test_basic.py::test_background_tasks_with_max_calls
+    // with -c opt on Windows.
+    RAY_CHECK(instance_->global_worker_) << "global_worker_ must not be NULL";
     return *instance_->global_worker_;
   }
   auto ptr = current_core_worker_.lock();

src/ray/gcs/asio.cc

@@ -79,7 +79,8 @@ void RedisAsioClient::operate() {
 
 void RedisAsioClient::handle_read(boost::system::error_code error_code) {
   RAY_CHECK(!error_code || error_code == boost::asio::error::would_block ||
-            error_code == boost::asio::error::connection_reset);
+            error_code == boost::asio::error::connection_reset)
+      << "handle_read(error_code = " << error_code << ")";
   read_in_progress_ = false;
   redis_async_context_.RedisAsyncHandleRead();
 
@@ -90,7 +91,8 @@ void RedisAsioClient::handle_read(boost::system::error_code error_code) {
 
 void RedisAsioClient::handle_write(boost::system::error_code error_code) {
   RAY_CHECK(!error_code || error_code == boost::asio::error::would_block ||
-            error_code == boost::asio::error::connection_reset);
+            error_code == boost::asio::error::connection_reset)
+      << "handle_write(error_code = " << error_code << ")";
   write_in_progress_ = false;
   redis_async_context_.RedisAsyncHandleWrite();
 

src/ray/gcs/redis_async_context.cc

@@ -50,7 +50,11 @@ void RedisAsyncContext::RedisAsyncHandleRead() {
   // This function will execute the callbacks which are registered by
   // `redisvAsyncCommand`, `redisAsyncCommandArgv` and so on.
   std::lock_guard<std::mutex> lock(mutex_);
-
+  // TODO(mehrdadn): Remove this when the bug is resolved.
+  // Somewhat consistently reproducible via
+  // python/ray/tests/test_basic.py::test_background_tasks_with_max_calls
+  // with -c opt on Windows.
+  RAY_CHECK(redis_async_context_) << "redis_async_context_ must not be NULL here";
   redisAsyncHandleRead(redis_async_context_);
 }