Cloudera have submitted a Software Grant Agreement for Impala to the ASF, clearing the way for us to contribute Impala's source code to git.apache.org.
Impala's repository is already set up. See https://git-wip-us.apache.org/repos/asf?p=incubator-impala.git;a=summary (and for general git@apache information, see https://git-wip-us.apache.org/).
Since Impala's repository is very large (over 10GB), we should take steps to reduce its size before that initial commit. The most significant consumer of space is /thirdparty; per recent discussions on the mailing list (see https://mail-archives.apache.org/mod_mbox/incubator-impala-dev/201603.mbox/browser) I propose we remove all traces of /thirdparty from the repository and commit the resulting branch.
I did this for a recent snapshot of Impala by filtering the branch and executing rm -rf /thirdparty on every commit. The result is at https://github.com/henryr/minimised-impala. The repository is now ~250MB which is much more reasonable. We should use that as a base and just catch the branch up to the most recent cdh5-trunk.
It's ok that this branch won't be buildable in the first instance. That's the subject of a subsequent JIRA.
Until development officially moves to git.apache.org (i.e. commits go there first, see the parent JIRA for steps), someone will need to periodically keep git.apache.org up-to-date. This should be scriptable.
Finally - the git.apache.org branch obviously should not be called cdh5-trunk. Let's just call it trunk instead to mimic other ASF projects.