当多个客户端推送到同一个远程存储库时,git如何处理并发问题? [英] When multiple clients push to same remote repository, how does git deal with concurrency issues?
问题描述
对于每个http请求,在服务器端运行单独的进程 git-http-backend
。
当多个客户端推送到相同的远程存储库时,它如何处理并发问题?
例如,它是否创建锁定文件和名称是什么? >
注意:远程仓库是裸的。
通过SSH调用 - 请参阅 refs.c中的update_ref函数。锁定机制的实际实现位于 lock_file函数在lockfile.c ,是的,它创建一个 .lock
文件。下面,通过 open()
的 O_EXCL
选项来完成。
For each http request, a separate process git-http-backend
runs on the server side.
When multiple clients push to same remote repository, what does it do to handle concurrency issues?
For example, does it create lock file and what the name is?
Note: Remote repository is bare.
In the same manner as if invoked via SSH -- see the update_ref function in refs.c. The actual implementation of the locking mechanism lives in the lock_file function in lockfile.c, and yes, it creates a .lock
file in there. Under the hood, it's done via the O_EXCL
option to open()
.
这篇关于当多个客户端推送到同一个远程存储库时,git如何处理并发问题?的文章就介绍到这了,希望我们推荐的答案对大家有所帮助,也希望大家多多支持IT屋!