On 01/11/2017 04:53 PM, Ian Wienand wrote: > The thing is, ansible 2.0.2.0 seems to do all the ssh stuff very > differently so this doesn't appear to happen. I tell a lie; this same thing actually happens with 2.0.2.0 I'm wondering if just nobody has run "run_all.sh" by hand (on a terminal) since [1]. I know we've frequently run the playbooks individually by hand, but not under timeout like that. After some more searching I've proposed [2] At the bottom of all this yak shaving, I think it's fine to upgrade puppetmaster to 2.2.1-rc3 to work around any potential for the recent CVE. I've proposed [3] and [4] for that. -i [1] https://review.openstack.org/#/c/311913/ [2] https://review.openstack.org/418722 Run timeout with --foreground [3] https://review.openstack.org/418652 [4] https://review.openstack.org/418671