Git push over HTTP not activating remote hooks

Question

On my remote box, I've initialized a bare git repository. In the hooks directory, I've initialized the post-receive, post-update and update hooks with the following script:

#!/bin/bash
echo $0 $@ >> /tmp/githooks.log

On my local box, I've cloned the repository, added a test file, committed it and pushed the change back to the remote box.

$ git clone https://remote/git/sandbox.git sandbox
$ cd sandbox
$ touch asdf
$ git add asdf
$ git commit -a
[master (root-commit) 37505de] zxcv
 0 files changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 asdf
$ git push origin master
Fetching remote heads...
  refs/
  refs/heads/
  refs/tags/
updating 'refs/heads/master'
  from 0000000000000000000000000000000000000000
  to   37505de9c22b0aee84e0071190f4f58728770675
    sending 3 objects
    done
Updating remote server info
To https://remote/git/sandbox.git
 * [new branch]      master -> master

However, /tmp/githooks.log is empty on the remote machine. If, however, I clone the repository while on the remote machine, the hooks are successfully called.

Do git hooks not work with http-push?

Answer 1

With Git protocols, you will have different features enabled.
For HTTP, this thread summarizes the issue:

The "problem" here (which is very much the way HTTP protocol was designed) is that it isn't git that updates repository on remote side on push (which knows about hooks), but web server via WebDAV.
And web server knows nothing about hooks.

Perhaps that would get improved when "smart" HTTP protocol gets implemented (currently in the phase of design, I think just after designing protocol).

As you commented, smart http would be the answer.

This feature is referred to as “smart” HTTP vs “dumb” HTTP because it requires having the Git binary installed on the server, where the previous incantation of HTTP transfer required only a simple webserver.
It has a real conversation with the client, rather than just dumbly pushing out data.