Backup_Slave/Backup_Master *base* state data structure
Logs parsing for Backup Slave and Backup Master are virtually the same, so the logic has been moved to shared parent - Backup.pm.
But the data is stored separately in state->{'Backup_Slave'} and state->{'Backup_Master'} depending on where it came from.
In all examples below, I write state->{'Backup*'}, and it relates to both places in state.
Within state->{'Backup*'} data is kept using following structure:
state->{'Backup*'}->[ n ] = { DATA }
Where
n is just a number, irrelevant. The only important fact is that higher n means that backup started later.
DATA - data about archiving this segment
DATA is arrayref which contains:
[0] - epoch of when omnipitr-backup was called
[1] - process number for omnipitr-backup-* program (pid)
[2] - epoch when backup was fully done.