search

aiidateam / aiida-core

The official repository for the AiiDA code
https://aiida-core.readthedocs.io
Other
435 stars 189 forks source link

verdi computer enable/disabled ignored #1550

Open giovannipizzi opened 6 years ago

giovannipizzi commented 6 years ago

Now, even if I run verdi computer disable <COMPUTERNAME> and then I submit there, the simulation is submitted anyway. Instead, no remote connection should be done while the computer is disabled (also for local transports).

Maybe check also if you want to solve #1154 at the same time

sphuber commented 6 years ago

What do you want the behavior to be? Should we let the code go to the submission attempt and check in there for the status of the computer and stop if it is disabled? Should this put the calculation into SUBMISSIONFAILED? Or should we do this further upstream and prevent the calculation from even being created and emitting a warning?

giovannipizzi commented 6 years ago

If we want to maintain the old behaviour, there should be an if check right before trying to ssh/connect to the machine, and just silently skip all calculations there if the computer is disabled (note: check the old logic, one should check both the global disabled state, and the one in the uauthinfo for the current user). One should also skip (or maybe in that case complain?) if the computer exists but it's not configured (again, the logic was there earlier, I believe). This is e.g. the case when you submit to a code that you imported.

As a bonus, I would also add a message to the bottom warnings (where we e.g. say that the daemon is off) if there is at least a calculation on a disabled computer).