Update retry policy

When an update is interrupted by a power outage, it is our observation that the deployment is marked as failed on the server side and the client does not reattempt to install it. This was also observed when the interruption occurred while the update was being downloaded. After that, the only way of re-trying to update to that particular version was to re-deploy on the server side.

Is it possible for clients to automatically re-try to apply failed updates? Or at least resume the download when interrupted?

Hi @dreamtripper yes, that is the current implementation. If the mender client restarts (whether due to a reboot or just “systemctl restart”) we have no means to understand if it is a result of a bad update or not. I know there was some talk about implementing automatic retry if we can determine that it is appropriate but there is no definite plan at the moment.


Thanks a lot! :slight_smile:

Is it possible to retrigger the update from the client side? Can it be done on the command line or using DBus or something similar?

All deployments are controlled by the server; the only way to trigger it from the client side would be to use the server API to create a new deployment.