From the discussion in the office: to guarantee the same state of dc0 and other domain controllers, it takes 2 osync passes (which runs crond once in samba_sysvol_sync_timeout minutes, and if it = 5, then we get a wait of at least 10 minutes after starting).

Из обсуждения в офисе: для гарантировано одинакового состояния dc0 и других контролеров домена, требуется ожидание 2 проходов osync (который запускается crond раз в samba_sysvol_sync_timeout минут, и если он = 5, то получаем ожидание минимум 10 минут после запуска).

I suggest to finish the task "AS IS" if it is working because we may cause scope creep. We may plan osync improvements for the future releases.

@mastersin accept pull request?

JFYI, As I can remember there were some issues with an initial data synchronization via osync among controllers. In some cases an actual data can be deleted on dc0 due an unexistance same data on rest of controllers. Such behaviour has prevented to merge previous PR and needs to be more investigated. I strongly recommends to start with rsync for purposes of SysVol replication and move to use the osync only in case of lack of functionality from rsync side.

osync_v1.2.pdf
Документация по osync от автора