100% progress bar during reinstall on primary-controller

[Veronica Krayneva]

100% progress bar during reinstall on primary-controller

287 ISO

Scenario:
1. Create cluster with neutron vlan
2. Add 3 controller and mongo roles
3. Add compute and cinder roles
4. Deploy cluster
5. Reinstall primary controller (provision + deploy)
6. Run OSTF

Cluster was deployed, but we can see next situation: progress bar is 100% during ~25-30 min
22 | running | deployment | 1 | 100 | 37069455-9aa4-4ca7-a655-d2fea1705f4e

[Ihor Kalnytskyi (ikalnytskyi)]

We don't fix Medium bugs for 7.0, so mark it as Won't fix and move it to 8.0

[Ihor Kalnytskyi (ikalnytskyi)]

Ok, I took a look and found out several issues.

1/ The provisioning shows the following error -

   Can't read file with logs: /var/log/remote/node-1.test.domain.local/bootstrap/fuel-agent.log

2/ We already have IBP image so we don't build it again and send to Nailgun "progress: 80" at the very beginning of provisioning.

3/ We reboot our node (looks like before provisioning), the timeout is expired - it took more that 240 seconds (4 minutes). Should be investigated and fixed.

4/ AS FOR ORIGINAL ISSUE

Astute reports the following progressed to Nailgun - 0, 0, 0, 0 ... 0, 7, 50, 50, 50, 100

Which is something wrong, since we report 100% at the middle of deployment.

[Vladimir Sharshov (vsharshov)]

Covered-by-bp: https://blueprints.launchpad.net/fuel/+spec/progress-bar-based-on-tasks

[Alexander Kislitsky (akislitsky)]

We passed SCF in 8.0. Moving the bug to 9.0.

[Vladimir Sharshov (vsharshov)]

Moving to Task Based Deployment solve the problem. Now progress bar calculated based on tasks which run on nodes.

[Alisa Tselovalnikova (atselovalnikova)]

Verified on http://paste.openstack.org/show/521126.