This is an automated email from the ASF dual-hosted git repository.
jamesge pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-brpc.git
The following commit(s) were added to refs/heads/master by this push:
new 1a3bfb9 Add docs on fork w/o exec
1a3bfb9 is described below
commit 1a3bfb97e410d65e525bc68bdd1e79e32476ed27
Author: Ge Jun <jge...@gmail.com>
AuthorDate: Fri Nov 15 10:18:06 2019 +0000
Add docs on fork w/o exec
---
docs/cn/server.md | 11 +++++++++++
docs/en/server.md | 11 +++++++++++
2 files changed, 22 insertions(+)
diff --git a/docs/cn/server.md b/docs/cn/server.md
index a2262c6..9f51987 100644
--- a/docs/cn/server.md
+++ b/docs/cn/server.md
@@ -341,6 +341,17 @@ server端会自动尝试其支持的协议，无需用户指定。`cntl->protoco
如果你有更多的协议需求，可以联系我们。
+# fork without exec
+一般来说，[fork](https://linux.die.net/man/3/fork)出的子进程应尽快调用[exec](https://linux.die.net/man/3/exec)以重置所有状态，中间只应调用满足async-signal-safe的函数。这么使用fork的brpc程序在之前的版本也不会有问题。
+
+但在一些场景中，用户想直接运行fork出的子进程，而不调用exec。由于fork只复制其调用者的线程，其余线程便随之消失了。对应到brpc中，bvar会依赖一个sampling_thread采样各种信息，在fork后便消失了，现象是很多bvar归零。
+
+最新版本的brpc会在fork后重建这个线程(如有必要)，从而使bvar在fork后能正常工作，再次fork也可以。已知问题是fork后cpu
profiler不正常。然而，这并不意味着用户可随意地fork，不管是brpc还是上层应用都会大量地创建线程，它们在fork后不会被重建，因为：
+* 大部分fork会紧接exec，浪费了重建
+* 给代码编写带来很多的麻烦和复杂度
+
+brpc的策略是按需创建这类线程，同时fork without
exec必须发生在所有可能创建这些线程的代码前。具体地说，至少**发生在初始化所有Server/Channel/应用代码前**，越早越好，不遵守这个约定的fork会导致程序不正常。另外，不支持fork
without exec的lib相当普遍，最好还是避免这种用法。
+
# 设置
## 版本
diff --git a/docs/en/server.md b/docs/en/server.md
index d604c1f..f28fd96 100644
--- a/docs/en/server.md
+++ b/docs/en/server.md
@@ -344,6 +344,17 @@ Server detects supported protocols automatically, without
assignment from users.
If you need more protocols, contact us.
+# fork without exec
+In general, [forked](https://linux.die.net/man/3/fork) subprocess should call
[exec](https://linux.die.net/man/3/exec) ASAP, before which only
async-signal-safe functions should be called. brpc programs using fork like
this should work correctly even in previous versions.
+
+But in some scenarios, users continue the subprocess without exec. Since fork
only copies its caller's thread, which causes other threads to disappear after
fork. In the case of brpc, bvar depends on a sampling_thread to sample various
information, which disappears after fork and causes many bvars to be zeros.
+
+Latest brpc re-creates the thread after fork(when necessary) to make bvar work
correctly, and can be forked again. A known problem is that the cpu profiler
does not work after fork. However users still can't call fork at any time,
since brpc and its applications create threads extensively, which are not
re-created after fork:
+* most fork continues with exec, which wastes re-creations
+* bring too many troubles and complexities to the code
+
+brpc's strategy is to create these threads on demand and fork without exec
should happen before all code that may create the threads. Specifically, **fork
without exec should happen before initializing all
Servers/Channels/Applications, earlier is better. fork not obeying this causes
the program dysfunctional. BTW, fork without exec better be avoided because
many libraries do not support it.
+
# Settings
## Version
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@brpc.apache.org
For additional commands, e-mail: dev-h...@brpc.apache.org
