We can do this via settings in /etc/cloud/cloud.cfg. E.g, for a
maximum wait time of 30 seconds:

datasource:
  Ec2:
    timeout: 10
    max_wait: 30

The default is a maximum wait time of 120 seconds.  The defaults come
from sources/DataSourceEc2.py:

    def _get_url_settings(self):
        mcfg = self.ds_cfg
        if not mcfg:
            mcfg = {}
        max_wait = 120
        try:
            max_wait = int(mcfg.get("max_wait", max_wait))
        except Exception:
            util.logexc(LOG, "Failed to get max wait. using %s", max_wait)

        if max_wait == 0:
            return False

        timeout = 50
        try:
            timeout = int(mcfg.get("timeout", timeout))
        except Exception:
            util.logexc(LOG, "Failed to get timeout, using %s", timeout)

        return (max_wait, timeout)

Does putting the above (timeout:10, max_wait: 30) in /etc/cloud/cloud.cfg make sense? It looks like this should work for at least anything as recent as 0.6.3.

Arg, one needs to set this *per datasource*.

We could add an alternate boot menu item that would append the following to the kernel command line:

    cc: datasource_list: [None] end_cc

This effectively disables cloud-init.

We do not have the resources to update the cloud-init package in RHEL 6.x at this time.