Improve error handling in Lancium adapter

[giffels]

During testing the integration of Lancium Compute into the infrastructure, TARDIS sometimes crashes due to two reasons, which can be internally handled by TARDIS.

1. Since we are updating the resource status only once a minute the following situation can occur. A drone previously in state queued or running can be changed meantime to error or finished. If TARDIS tries to terminate this drone, the Lancium API returns HTTP Status Code 419 ("Unable to terminate a job that is not queued or running"). In that case TARDIS should handle it as TardisResourceStatusUpdateFailed exception. So, after the next resource status update, the life cycle management can take care of it.
2. It can rarely happen that a drone crashes and is not anymore in the system, so that the Lancium API returns a HTTP Status Code 404 ("Not found"). In that case TARDISshould handle it as TardisDroneCrashed exception. Afterwards the life cycle management takes care of it.

FYI the handle_exceptions context manager is involved in here.

[RHofsaess]

I think it looks fine 👍

[eileen-kuehn]

2. It can rarely happen that a drone crashes and is not anymore in the system, so that the Lancium API returns a HTTP Status Code 404 ("Not found"). In that case TARDISshould handle it as TardisDroneCrashed exception. Afterwards the life cycle management takes care of it.

I have a quick question on this. Does the handling of a drone as crashed have any further meaning? It can (or should) happen that a HTTP Status Code 404 is given even when the drone has not crashed but was correctly terminated by TARDIS (or something else), e.g. the confirmation might just have gone lost.

In case it does not make any difference, I will approve :)

[giffels]

2. It can rarely happen that a drone crashes and is not anymore in the system, so that the Lancium API returns a HTTP Status Code 404 ("Not found"). In that case TARDISshould handle it as TardisDroneCrashed exception. Afterwards the life cycle management takes care of it.

I have a quick question on this. Does the handling of a drone as crashed have any further meaning?

The drone is not handled anymore by TARDIS, since it is already gone.

It can (or should) happen that a HTTP Status Code 404 is given even when the drone has not crashed but was correctly terminated by TARDIS (or something else), e.g. the confirmation might just have gone lost.

Yes, that can happen, however the result is the same. The drone is gone and it is not handled anymore by TARDIS.

In case it does not make any difference, I will approve :)

Yes, does not make any difference! :-)

[RHofsaess]

Still fine for me :D