This is an automated email from the ASF dual-hosted git repository.
jiashunzhu pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/brpc.git
The following commit(s) were added to refs/heads/master by this push:
new a665f60d remove the docs and comments related to HealthCheckThread
(#2463)
a665f60d is described below
commit a665f60d10eb1fd991941def0707b872231e4a09
Author: Divyansh200102 <146909065+divyansh200...@users.noreply.github.com>
AuthorDate: Thu Dec 14 03:23:48 2023 +0530
remove the docs and comments related to HealthCheckThread (#2463)
---
docs/cn/load_balancing.md | 2 +-
src/brpc/socket.h | 2 +-
test/brpc_socket_unittest.cpp | 6 +++---
3 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/docs/cn/load_balancing.md b/docs/cn/load_balancing.md
index 27bad7d3..689c1e8c 100644
--- a/docs/cn/load_balancing.md
+++ b/docs/cn/load_balancing.md
@@ -40,7 +40,7 @@ Load balancer最重要的是如何让不同线程中的负载均衡不互斥，
对于那些无法连接却仍在NamingService的节点，brpc会定期连接它们，成功后对应的Socket将被”复活“，并可能被LoadBalancer选择上，这个过程就是健康检查。注意：被健康检查或在LoadBalancer中的节点一定在NamingService中。换句话说，只要一个节点不从NamingService删除，它要么是正常的（会被LoadBalancer选上），要么在做健康检查。
-传统的做法是使用一个线程做所有连接的健康检查，brpc简化了这个过程：为需要的连接动态创建一个bthread专门做健康检查（Socket::HealthCheckThread）。这个线程的生命周期被对应连接管理。具体来说，当Socket被SetFailed后，健康检查线程就可能启动（如果SocketOptions.health_check_interval为正数的话）：
+传统的做法是使用一个线程做所有连接的健康检查，brpc简化了这个过程：为需要的连接动态创建一个bthread专门做健康检查（Socket::StartHealthCheck）。这个线程的生命周期被对应连接管理。具体来说，当Socket被SetFailed后，健康检查线程就可能启动（如果SocketOptions.health_check_interval为正数的话）：
-
健康检查线程先在确保没有其他人在使用Socket了后关闭连接。目前是通过对Socket的引用计数判断的。这个方法之所以有效在于Socket被SetFailed后就不能被Address了，所以引用计数只减不增。
- 定期连接直到远端机器被连接上，在这个过程中，如果Socket析构了，那么该线程也就随之退出了。
diff --git a/src/brpc/socket.h b/src/brpc/socket.h
index f4c3cf90..d28d8f17 100644
--- a/src/brpc/socket.h
+++ b/src/brpc/socket.h
@@ -386,7 +386,7 @@ public:
// Mark this Socket or the Socket associated with `id' as failed.
// Any later Address() of the identifier shall return NULL unless the
- // Socket was revivied by HealthCheckThread. The Socket is NOT recycled
+ // Socket was revivied by StartHealthCheck. The Socket is NOT recycled
// after calling this function, instead it will be recycled when no one
// references it. Internal fields of the Socket are still accessible
// after calling this function. Calling SetFailed() of a Socket more
diff --git a/test/brpc_socket_unittest.cpp b/test/brpc_socket_unittest.cpp
index 36a3b1b0..d2258735 100644
--- a/test/brpc_socket_unittest.cpp
+++ b/test/brpc_socket_unittest.cpp
@@ -543,7 +543,7 @@ TEST_F(SocketTest, not_health_check_when_nref_hits_0) {
ASSERT_TRUE(src.empty());
ASSERT_EQ(-1, s->fd());
}
- // HealthCheckThread is possibly still running. Spin until global_sock
+ // StartHealthCheck is possibly still running. Spin until global_sock
// is NULL(set in CheckRecycle::BeforeRecycle). Notice that you should
// not spin until Socket::Status(id) becomes -1 and assert global_sock
// to be NULL because invalidating id happens before calling BeforeRecycle.
@@ -765,7 +765,7 @@ TEST_F(SocketTest, health_check) {
s.release()->Dereference();
- // Must stop messenger before SetFailed the id otherwise HealthCheckThread
+ // Must stop messenger before SetFailed the id otherwise StartHealthCheck
// still has chance to get reconnected and revive the id.
messenger->StopAccept(0);
ASSERT_EQ(-1, messenger->listened_fd());
@@ -773,7 +773,7 @@ TEST_F(SocketTest, health_check) {
ASSERT_EQ(EBADF, errno);
ASSERT_EQ(0, brpc::Socket::SetFailed(id));
- // HealthCheckThread is possibly still addressing the Socket.
+ // StartHealthCheck is possibly still addressing the Socket.
start_time = butil::gettimeofday_us();
while (global_sock != NULL) {
bthread_usleep(1000);
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@brpc.apache.org
For additional commands, e-mail: dev-h...@brpc.apache.org
