During upgrare M->N ssh key population doesn't work

[Ionuț Bîru]

I've got the following fatal error:

fatal: [cloudbox1]: FAILED! => {"failed": true, "msg": "the file_name '/var/tmp/openstack-nova-key.sh' does not exist, or is not readable"}

I was thinking that maybe it's because I do have a broken compute node that is not reachable, since I saw that a previous bug was fixed related to that.

I did skip that particular compute node with --limit '!cloudbox2' but didn't resolve the problem.

What worked is to revert this commit:
https://git.openstack.org/cgit/openstack/openstack-ansible-os_nova/commit/tasks/nova_compute_key_distribute.yml?id=bda35e7fd60b48af24b0b7501bffba151ab2d187

Log here:
https://paste.xinu.at/17Zz/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/385221
Committed: https://git.openstack.org/cgit/openstack/openstack-ansible-os_nova/commit/?id=3f91e25a0a427db45d9919f4624d0dc0f6ef3dd0
Submitter: Jenkins
Branch: master

commit 3f91e25a0a427db45d9919f4624d0dc0f6ef3dd0
Author: Jimmy McCrory <email address hidden>
Date: Tue Oct 11 17:57:28 2016 -0700

    Delegate from first available compute host

    The 'Drop authorized keys file script locally' task shouldn't
    specifically require that the first member of the nova_compute group be
    available. If the playbook is run with a limit excluding that host, the
    task will always end up being skipped.

    Use an intersect filter to find the first available compute host within
    the play and use that as the delegation host.

    Closes-Bug: 1632447
    Change-Id: Id12fd6cee104a6fb705acc66704f21d629694d84

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/newton
Review: https://review.openstack.org/385991

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/385991
Committed: https://git.openstack.org/cgit/openstack/openstack-ansible-os_nova/commit/?id=4813e76b929b18d8aa6cf7292581cfc2316b69b1
Submitter: Jenkins
Branch: stable/newton

commit 4813e76b929b18d8aa6cf7292581cfc2316b69b1
Author: Jimmy McCrory <email address hidden>
Date: Tue Oct 11 17:57:28 2016 -0700

    Delegate from first available compute host

    The 'Drop authorized keys file script locally' task shouldn't
    specifically require that the first member of the nova_compute group be
    available. If the playbook is run with a limit excluding that host, the
    task will always end up being skipped.

    Use an intersect filter to find the first available compute host within
    the play and use that as the delegation host.

    Closes-Bug: 1632447
    Change-Id: Id12fd6cee104a6fb705acc66704f21d629694d84
    (cherry picked from commit 3f91e25a0a427db45d9919f4624d0dc0f6ef3dd0)

[Ionuț Bîru (ionut-3)]

This bug is not resolved with the above commit.

Attaching the log

[Jimmy McCrory (jimmy-mccrory)]

It looks like your cloudbox2 host is still unreachable.
Could you try running the playbook again with a limit to exclude it?

The task is going to find and try to use that host if it's included in the play.

[Ionuț Bîru (ionut-3)]

It works fine if I skip that compute host.

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/openstack-ansible-os_nova 14.0.0.0rc4 release candidate.

[Ionuț Bîru (ionut-3)]

The proposed fix is indeed available in openstack/openstack-ansible-os_nova 14.0.0.0rc4 .
What I'm saying is that the fix doesn't resolve this bug.

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/openstack-ansible-os_nova 15.0.0.0b1 development milestone.