我建了一个GitLab服务器,操作系统是Mac OS X 10.9.2,上边有两个仓库,之前两个仓库都可以正常clone,但是最近两天突然发现有一个仓库clone时,在本地机上遇到:
Cloning into 'gamil'...
remote: Counting objects: 426, done.
remote: Compressing objects: 100% (375/375), done.
fatal: The remote end hung up unexpectedly
fatal: early EOF
fatal: index-pack failed
然后我检查了GitLab里的日志unicorn.stderr.log
,发现
error: git-upload-pack died of signal 13
这个仓库并不大,只有2M多。我查看了信号13的意义:
terminate process write on a pipe with no reader
请问如何解决呢?
The problem is caused by the nginx tmp directory not having permissions for the adim group. My system is Mac OS X, and I use nginx and other software installed with Homebrew. I had always suspected it was a timeout issue, but I didn't find that a certain Homebrew update caused a tmp permissions issue.
In addition, for smaller warehouses, it seems that there is no need to create files under tmp, so one warehouse can be cloned, but the other cannot.
Does this phenomenon happen repeatedly?
Can I clone successfully using SSH directly without using Gitlab?
Can shallow clone and pull be successful?
There is a saying that it is caused by the memory of the git server being too small: http://stackoverflow.com/questions/1943914/git-clone-fails-with-index-pack-failed