DevStack "stack.sh" sometimes fails after "unstack.sh"

[Ilya Chukhnakov]

Steps to reproduce:
1. stack.sh (successful)
2. restart host
3. unstack.sh
4. stack.sh (failing)

Workaround:
Run unstack.sh+stack.sh again after failure occurs.

DevStack output:
2016-11-03 11:30:25.436 | ++ /opt/stack/kuryr-kubernetes/devstack/plugin.sh:run_etcd:121 : run_container etcd --net host --volume=/var/lib/
etcd:/var/etcd:rw quay.io/coreos/etcd:v3.0.8 /usr/local/bin/etcd --name devstack --data-dir /var/etcd/data --initial-advertise-peer-urls http
://172.16.240.11:2380 --listen-peer-urls http://0.0.0.0:2380 --listen-client-urls http://0.0.0.0:2379 --advertise-client-urls http://172.16.2
40.11:2379 --initial-cluster-token etcd-cluster-1 --initial-cluster devstack=http://172.16.240.11:2380 --initial-cluster-state new
2016-11-03 11:30:25.441 | ++ /opt/stack/kuryr-kubernetes/devstack/plugin.sh:run_container:17 : local name
2016-11-03 11:30:25.442 | ++ /opt/stack/kuryr-kubernetes/devstack/plugin.sh:run_container:19 : name=etcd
2016-11-03 11:30:25.444 | ++ /opt/stack/kuryr-kubernetes/devstack/plugin.sh:run_container:20 : shift
2016-11-03 11:30:25.445 | ++ /opt/stack/kuryr-kubernetes/devstack/plugin.sh:run_container:22 : docker run --name etcd --detach --net host -
-volume=/var/lib/etcd:/var/etcd:rw quay.io/coreos/etcd:v3.0.8 /usr/local/bin/etcd --name devstack --data-dir /var/etcd/data --initial-adverti
se-peer-urls http://172.16.240.11:2380 --listen-peer-urls http://0.0.0.0:2380 --listen-client-urls http://0.0.0.0:2379 --advertise-client-url
s http://172.16.240.11:2379 --initial-cluster-token etcd-cluster-1 --initial-cluster devstack=http://172.16.240.11:2380 --initial-cluster-sta
te new
2016-11-03 11:30:25.453 | docker: Error response from daemon: Conflict. The name "/etcd" is already in use by container b77530f9463a075a2816e
c90af27776ec2d2486fd6f5185c46d0afeac3117c66. You have to remove (or rename) that container to be able to reuse that name..
2016-11-03 11:30:25.456 | See 'docker run --help'.

[Luis Tomas Bolivar (ltomasbo)]

Yes, I hit this one too. Just by removing the docker etcd container (docker rm etcd_id) it works

[Daniel Mellado (daniel-mellado)]

I'll investigate it, the code should be doing exactly that...

[Antoni Segura Puimedon (celebdor)]

We'll solve it by moving to devstack's etcd plugin

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/474211

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: master
Review: https://review.openstack.org/475036

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/474211
Committed: https://git.openstack.org/cgit/openstack/kuryr-kubernetes/commit/?id=9500a81a57383017eec298a55d343faa446d9974
Submitter: Jenkins
Branch: master

commit 9500a81a57383017eec298a55d343faa446d9974
Author: Antoni Segura Puimedon <email address hidden>
Date: Wed Jun 14 16:01:08 2017 +0200

    devstack: Use devstack's etcd3 service

    This patch deprecates the containerized etcd that we've been using up
    until now for devstack. It moves the default to be the devstack provided
    etcd3 service so we do not conflict with other etcd3 consumers and
    benefit from the upstream devstack improvements.

    Change-Id: Icf7d48b72d0ff6ae34d4752040100997d7892c5e
    Fixes-Bug: 1638892
    Signed-off-by: Antoni Segura Puimedon <email address hidden>

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/475036
Committed: https://git.openstack.org/cgit/openstack/kuryr-kubernetes/commit/?id=976d7af0a8655850baa1d21336dcf60aad11288b
Submitter: Jenkins
Branch: master

commit 976d7af0a8655850baa1d21336dcf60aad11288b
Author: Kirill Zaitsev <email address hidden>
Date: Fri Jun 16 20:27:48 2017 +0300

    Switch gate hook to use etcd3 instead of legacy

    Related-Bug: #1638892
    Change-Id: I306b37630d0088619dcf5ea75662452c4de2eb48