I have made update and rollback for CentOS and Ubuntu clusters.
Steps to reproduce:
1. Create env with Fuel 5.0.26 (2014.1)
2. Create CentOS cluster: (simple, neutron/vlan, ceilometer, ceph vol/ceph images, 1 controller+mongo, 4 compute+cinder+ceph)
3. Store packages versions from cluster nodes to a file (5.0.0.26.orig.sorted.txt for centos and 5.0.0.26-before-update.sorted.txt for ubuntu)
4. Upgrade env to 5.1 using http://jenkins-product.srt.mirantis.net:8080/job/fuel_master_iso/493/
5. Update CentOS cluster to 5.0.2
6. Store packages versions from cluster nodes to a file (5.0.2.versions.sorted.txt for centos and 5.0.2-updated.sorted.txt for ubuntu)
7. Run OSTF
8. Make diagnostic snapshot (centos-5.0.2-fuel-snapshot and ubuntu-5.0.2-fuel-snapshot was made)
9. Rollback cluster to 5.0.0.26
10. Store packages versions from cluster nodes to a file (5.0.0.26-reverted.versions.sorted.txt for centos and 5.0.0.26-reverted.sorted.txt for ubuntu)
11. Make diagnostic snapshot (centos-reverted-to-5.0.0.26 and ubuntu-reverted-to-5.0.0.26)
The following are diagnostic snapshots:
centos-5.0.2-fuel-snapshot-2014-09-09_09-13-41.tgz
ubuntu-5.0.2-fuel-snapshot-2014-09-03_14-06-18.tgz
centos-reverted-to-5.0.0.26~ot-2014-09-09_11-43-48.tgz
ubuntu-reverted-to-5.0.0.26~ot-2014-09-03_14-06-18.tgz
The following are packages lists before updating cluster, after updating and after rollback (there is lists for controller and just one of compute nodes since they are the same):
centos-reverted-5.0.0.26.tar.gz
ubuntu-reverted-5.0.0.26.tar.gz
P.S. Updating of CentOS cluster to 5.0.2 failed, it is described in the https://bugs.launchpad.net/fuel/+bug/1367158 . Rollback for CentOS failed too. So on compute nodes some packages were not downgraded.
I have made update and rollback for CentOS and Ubuntu clusters.
Steps to reproduce: cinder+ ceph) 26.orig. sorted. txt for centos and 5.0.0.26- before- update. sorted. txt for ubuntu) jenkins- product. srt.mirantis. net:8080/ job/fuel_ master_ iso/493/ versions. sorted. txt for centos and 5.0.2-updated. sorted. txt for ubuntu) 5.0.2-fuel- snapshot and ubuntu- 5.0.2-fuel- snapshot was made) 26-reverted. versions. sorted. txt for centos and 5.0.0.26- reverted. sorted. txt for ubuntu) reverted- to-5.0. 0.26 and ubuntu- reverted- to-5.0. 0.26)
1. Create env with Fuel 5.0.26 (2014.1)
2. Create CentOS cluster: (simple, neutron/vlan, ceilometer, ceph vol/ceph images, 1 controller+mongo, 4 compute+
3. Store packages versions from cluster nodes to a file (5.0.0.
4. Upgrade env to 5.1 using http://
5. Update CentOS cluster to 5.0.2
6. Store packages versions from cluster nodes to a file (5.0.2.
7. Run OSTF
8. Make diagnostic snapshot (centos-
9. Rollback cluster to 5.0.0.26
10. Store packages versions from cluster nodes to a file (5.0.0.
11. Make diagnostic snapshot (centos-
The following are diagnostic snapshots: 5.0.2-fuel- snapshot- 2014-09- 09_09-13- 41.tgz 5.0.2-fuel- snapshot- 2014-09- 03_14-06- 18.tgz reverted- to-5.0. 0.26~ot- 2014-09- 09_11-43- 48.tgz reverted- to-5.0. 0.26~ot- 2014-09- 03_14-06- 18.tgz
centos-
ubuntu-
centos-
ubuntu-
The following are packages lists before updating cluster, after updating and after rollback (there is lists for controller and just one of compute nodes since they are the same): reverted- 5.0.0.26. tar.gz reverted- 5.0.0.26. tar.gz
centos-
ubuntu-
P.S. Updating of CentOS cluster to 5.0.2 failed, it is described in the https:/ /bugs.launchpad .net/fuel/ +bug/1367158 . Rollback for CentOS failed too. So on compute nodes some packages were not downgraded.