Which snapshot algorithm does zeebe use, like Chandy-Lamport?
The running state of the partition is captured periodically on the leader in a snapshot. By default, this period is every five minutes. This can be changed in the configuration.
A snapshot is a projection of all events that represent the current running state of the processes running on the partition. It contains all active data, for example, deployed processes, active process instances, and not yet completed jobs.
When the broker writes a new snapshot, it deletes all data on the log which was written before the latest snapshot.
I am not sure about the actual algorithm which is used. Maybe @menski has some more insights on this?
Hopefully my little “excursus” was already helpful for you.
So does this snapshot also include the running data of the microservice at that time?
You are welcome!
Data from microservices is not touched and included in the backup.