Comment 3 for bug 1496130

Revision history for this message
Stuart Bishop (stub) wrote : Re: 0.1.14 deployments eternally hanging

@thedac - charms cannot declare themselves ready, because they have no way of knowing when if there are hooks still scheduled to run, or when the operator (in this case, mojo) has stopped issuing commands.

For others who have had trouble with juju-wait, I'd love a bug report. I have *never* seen this algorithm fail in my tests, and have never had a report of it failing. I have only ever received a single bug report, and that was from running multiple 'juju run' commands in parallel (it confuses juju). I have heard of people suspicious about it, because it is taking a long time (and turns out, correctly waiting as the system is still busy running hooks despite it not being visible to the end user). Its been used in my amulet tests for ages, and I keep recommending it whenever I see charmers tripping over wait problems. So if it doesn't work, please tell somebody. Now we need to fix this bug twice (once in mojo, once in juju-wait).