Database left in an incorect state if interrupting an ongoing deployment

[PaulFarault]

An interrupted deployment keeps its status as RUNNING.

Is it a normal behavior? How to deal with this case as nothing failed but a few operations already succeeded?

[PaulFarault]

It seems normal to left the database "as is" in this case. What we should look for is how to recover from this state (maybe using a resume?).

[sergkudinov]

To my mind, if we have resume, it is naturally to have a pause command. In the context of TDP, it would be very useful.

Deployment process could be interrupted by the user in 2 manners:

• sending a temination signal with CTRL+C or clossing the parrent process
• running a command, for example tdp pause command or via tdp-server in the future.

With pause, we will need status PAUSED for deployments and operations. Having a detached mode from the terminal for the tdp deploy command would be also nice (option -d for example) . That would complement this current issue with a better usage.

@rpignolet what do you think?