[OSTF] OSTF processes should be killed when resetting or stopping deployment task starts

[Tatyanka]

Steps to reproduce:
1. Deploy cluster
2. When cluster is deployed successfully invoke ostf
3. When tests executions begin - go to the action tab and reset cluster
4. When reset task finish - back to health check tab and see results

Expected results:
there are no running tests on reset cluster. Also there is no forked running ostf processes on muster node

Actual Result:
Tests are running (some of them failed without any message (nodes are in bootrap))
Forked ostf proccess are running om master node

root 14973 0.1 5.0 342952 51648 ? S 13:14 0:03 \_ /opt/fuel_plugins/ostf/bin/python /opt/fuel_plugins/ostf/bin/ostf-server --host=127.0.0.1 --port=8777 --log_file=/var/log/ostf.log --dbpath postgresql+psycopg2://ostf:ostf@127.0.0.1:5432/ostf
root 17827 0.0 4.6 341920 47560 ? S 14:07 0:00 \_ /opt/fuel_plugins/ostf/bin/python /opt/fuel_plugins/ostf/bin/ostf-server --host=127.0.0.1 --port=8777 --log_file=/var/log/ostf.log --dbpath postgresql+psycopg2://ostf:ostf@127.0.0.1:5432/ostf

[Evgeniy L (rustyrobot)]

It should not be done on nailgun or orchestrator side.
Nailgun and orchestrator should know nothing about ostf and other fuel plugins.
Do you have some kind of master process where you can monitor state of cluster and in case if cluster changed its state just kill child processes?

[Tatyanka (tatyana-leontovich)]

need to be discussed, as for now, we do not have such process. May be we should add some polling service(in ostf) that will poll cluster status and ostf processes(include main process and it subproc) and interrupt it(subproc) by signal in case if cluster is not ready. Lets discuss it in irc

[Dima Shulyak (dshulyak)]

Evgeniy, ostf it is just client-server application.

To add process that monitors nailgun state - means adding separate process, that will be wasted in 99% time.

Good approach would be to use rabbitmq notifications, with fanout topic, where any application can listen for events.

As a quick fix, i dont see any issues if this will be implemented in astute.

[Dmitry Pyzhov (dpyzhov)]

We are lack of resources in 6.1. Moved to 7.0

[Dmitry Pyzhov (dpyzhov)]

It is a feature request and we cannot backport it to Mitaka.