Worker doesn't pick up the new workflow

[Cbkhare]

When a worker is rebooted via an action in the workflow, The worker picks up the workflow which is stuck in running state and as a result, it never picks up the new workflow.
This is in reference to the recent commit #115.

Expected Behaviour

Ideally, changes in this commit #115 should be rectified to handle the following scenario.

• tink-server should have a mechanism to execute/resume running workflow after reboot.
• When rebooted via workflow either workflow should be marked as completed.
• Or worker should pick up new workflow (if there is one defined for it).

Current Behaviour.

Below are the stepwise states happening right now.

• worker does ipxeboot.

• worker executes the workflow.

• worker reboots as a part of workflow.

• worker, after reboot, retries to pxe boot (because of hasActiveWorkflow should consider workflows in state PENDING and RUNNING #115 ) and fails.

• When a new workflow is created.

• Worker stills picks up the workflow which is running state and against, fails.

Steps to Reproduce (for bugs)

• Create a workflow using the below template.
• Use any alpine image for executing reboot action.

version: "0.1"
name: try-install
global_timeout: 1800
tasks:
  - name: "hello world"
    worker: "{{.device_1}}"
    actions:
      - name: "hello_world"
        image: hello-world
        timeout: 60
      - name: "reboot"
        image: alpine-image
        command:
          - bash
          - -c
          - |
            echo 1 > /proc/sys/kernel/sysrq; echo b > /proc/sysrq-trigger

Observe steps as mentioned in current behaviour.

• State of first workflow

docker exec -i deploy_tink-cli_1 tink workflow state 88241ed8-6221-11eb-af88-0242ac120005 
+----------------------+--------------------------------------+
| FIELD NAME           | VALUES                               |
+----------------------+--------------------------------------+
| Workflow ID          | 88241ed8-6221-11eb-af88-0242ac120005 |
| Workflow Progress    | 50%                                  |
| Current Task         | hello world                          |
| Current Action       | reboot                               |
| Current Worker       | 0eba0bf8-3772-4b4a-ab9f-6ebe93b90a95 |
| Current Action State | STATE_RUNNING                        |
+----------------------+--------------------------------------+

• state of the worker after reboot action.

• Create a new workflow.
• worker still picks up the old workflow because of hasActiveWorkflow should consider workflows in state PENDING and RUNNING #115 and fails again.
  
• Both of the workflow new and old, are stuck at their respective state.

vagrant@provisioner:/vagrant/deploy/centos$ docker exec -i deploy_tink-cli_1 tink workflow state 88241ed8-6221-11eb-af88-0242ac120005 
+----------------------+--------------------------------------+
| FIELD NAME           | VALUES                               |
+----------------------+--------------------------------------+
| Workflow ID          | 88241ed8-6221-11eb-af88-0242ac120005 |
| Workflow Progress    | 50%                                  |
| Current Task         | hello world                          |
| Current Action       | reboot                               |
| Current Worker       | 0eba0bf8-3772-4b4a-ab9f-6ebe93b90a95 |
| Current Action State | STATE_RUNNING                        |
+----------------------+--------------------------------------+
vagrant@provisioner:/vagrant/deploy/centos$ docker exec -i deploy_tink-cli_1 tink workflow state 5894621e-6222-11eb-af88-0242ac120005
+----------------------+--------------------------------------+
| FIELD NAME           | VALUES                               |
+----------------------+--------------------------------------+
| Workflow ID          | 5894621e-6222-11eb-af88-0242ac120005 |
| Workflow Progress    | 0%                                   |
| Current Task         |                                      |
| Current Action       |                                      |
| Current Worker       |                                      |
| Current Action State | STATE_PENDING                        |
+----------------------+--------------------------------------+

Result: If used for OS installation and worker is rebooted as a part of the workflow. The worker will never come up.

Context

• Creating a centos workflow.

Your Environment

• Vagrant

[mmlb]

We can't do anything about this at the moment I think. The way around this is to sleep for a little bit before doing the echo so that tink-worker has had the ability to send progress to tink-server.

[mmlb]

Hmm can you give boots logs for the attempted dhcp. It should allow the dhcp/tftp/...netboot.

[mmlb]

The action is the issue here, as written it doesn't allow for tink-server to report back success and so naturally we re-run it. This is working as intended. You'll want to use the reboot action from tinkerbell/actions#13 when its ready.