I have async-Classic file based replication with 2 Nodes as Active - Active. R/W is happening on Primary Node (PN) only and Secondary Node (SN) has been setup as HA only. SN has been configured as read_only mode.
In case if PN Database goes down when I/O thread (of SN) trying to fetch bin-logs resulting which not all events got replicated to SN. Now , data comes in unsync state .
As per my another script i will re-direct all R/W operations to SN by disabling its read_only mode post PN goes down.
Now question is :
Q. If i copy those binary logs from PN at point where replication got break and get it executed at SN end. how much it is feasible ? Things i Know : For this , can i use execute mysqldump utility from SN with its *read-binary-logs-from-server* parameter. This will copy the logs from PN to SN and then i can execute it . But my doubt is , as PN DB is down ( but VM box is running ), can i still be able to fetch bin-logs ?
Please give your feedback on this . How much this solution will work . I cannot use any other 3rd party tool or other modes of replication due to time constraints and cost constraints
If DB at PN is down , i cannot make use of mysqldump .
What i have figured out is , we can take the position at which replication went down and i will fetch those records from binary logs and get it executed at SN. In this way un-replicated records get inserted at SN