Did you try to clone with the http address?
The is currently a ticket with msysgit (issue 136) with the same problem, for which the current workaround is to clone with the http rather than the git address.
Other moves to try:
GIT_TRACE=1
(and post the output)
- try with other or older versions of msysgit
- check for any firewall/antivirus issue on your main computer
Here switching to an older version of msysgit could be the right option:
See this GitHub support thread:
Are you using the 1.6.4 prerelease of msysGit? There are known bugs with it.
We recommend a stable version, such as 1.6.0.
thanks chris, that's it: after switching back to the previous version of msysGit (Git-1.6.3.2-preview20090608.exe) the problem does not occur any more.
The problem might actually affect any Git1.6.4 and older:
git tries to check if the pack file exists by sending out a HEAD
request, but chokes on the 500 error that some (if not all) github returns.
Using the repository used by the reporter:
$ curl http://github.com/grails/grails.git/objects/info/packs
P pack-1290e84bed53bda28f0989dca48d836bd9104031.pack
P pack-bf40d38ae780512994e5127e832ed9d8853c186d.pack
P pack-f490d5f7d4671368f4a52c618ca9dce13b714ba1.pack
P pack-79e3a7f30e8989acc8403ac688be669a05384eef.pack
$ curl -I http://github.com/grails/grails.git/objects/pack/pack-bf40d38ae780512994e5127e832ed9d8853c186d.pack
HTTP/1.1 500 Internal Server Error
Server: nginx/0.6.26
Date: Fri, 04 Sep 2009 13:50:20 GMT
Content-Type: text/html; charset=utf-8
Connection: keep-alive
Content-Length: 3193
Cache-Control: no-cache
See this thread.
The GitHub support teams is currently investigating the problem:
I don't think the 500s are intentional, but there may be something odd with our setup that is causing them.
I have a ticket open to investigate the issue further. I have a feeling we won't be messing with this till after the move, since it's likely something with the server config.
Git1.6.4.4 does contain a fix since September, 16th:
The workaround for Github server that sometimes gave 500 (Internal server error) response to HEAD requests in 1.6.4.3 introduced a regression that caused re-fetching projects over http to segfault
in certain cases due to uninitialized pointer being freed.
But mssysgit has yet to release (September 21th) any new build since the original 1.6.4 late July. (They are probably waiting for 1.6.5 or 1.7)