You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the problem
On client shutdown or task suspend, non-checkpointed work is lost. So far, I've found uncheckpointed work for 1h or even 2h worth, which is quite a waste, if stopped and lost.
When an orderly shutdown is possible (eg. reviewing tasks before shutdown/restart), you can wait tasks to checkpoint before suspending/quitting, but it's tedious and inconvenient.
It can be helpful when shutting down, but also for switching tasks/projects.
Describe the solution you'd like
I would like to have a 'Suspend after next checkpoint' task action. Thus, I could send it to all running tasks to guarantee that no work is lost.
Additional context
Assumptions:
Tasks notify client on checkpoint, which can react accordingly.
Some tasks do not checkpoint, which is fine. It's up to you to know it if you intend to do "whatever at checkpoint".
This feature would later enable also a 'checkpointed shutdown' for the whole client. You can manually trigger it as a command when winding down before shutdown, set it as an opt-in behaviour on SIGTERM or do it on another signal. Lots of people might find it helpful to avoid wasted work (time and electricity).
The text was updated successfully, but these errors were encountered:
Describe the problem
On client shutdown or task suspend, non-checkpointed work is lost. So far, I've found uncheckpointed work for 1h or even 2h worth, which is quite a waste, if stopped and lost.
When an orderly shutdown is possible (eg. reviewing tasks before shutdown/restart), you can wait tasks to checkpoint before suspending/quitting, but it's tedious and inconvenient.
It can be helpful when shutting down, but also for switching tasks/projects.
Describe the solution you'd like
I would like to have a 'Suspend after next checkpoint' task action. Thus, I could send it to all running tasks to guarantee that no work is lost.
Additional context
Assumptions:
This feature would later enable also a 'checkpointed shutdown' for the whole client. You can manually trigger it as a command when winding down before shutdown, set it as an opt-in behaviour on
SIGTERM
or do it on another signal. Lots of people might find it helpful to avoid wasted work (time and electricity).The text was updated successfully, but these errors were encountered: