Re: How R/O node catch up global.gtid_executed of R/W master's?
Posted by:
Adam Nelson
Date: September 13, 2017 08:19AM
1) db02 should go into recovery mode until it is caught up. Until it has recovered it will not be marked as "ONLINE" in the cluster. Assuming that you are using mysql-router (or something to that effect to manage the queries) queries should not be directed to any member that is not online. Details can be found here :
https://dev.mysql.com/doc/refman/5.7/en/group-replication-basics.html
2) GTID is created for every transaction. Not just for every write operation. As long as there is query activity the GTIDs will increase. Only write transactions are replicated out to the group but every transaction gets a GTID. Details on the GTID can be found here :
https://dev.mysql.com/doc/refman/5.6/en/replication-gtids-concepts.html
3) I haven't tested this scenario. However, my assumption is that if you connected to it directly it would respond normally. However, the data would be out of date. It would be in a RO state and would reject any RW requests just like a normal RO node. That being said ... this is why you need either mysql-router or a similar product infont of the cluster managing queries or connections. Otherwise the application needs to be aware of the state of each node and would be responsible for directing the queries correctly.
Subject
Views
Written By
Posted
781
August 17, 2017 08:14PM
Re: How R/O node catch up global.gtid_executed of R/W master's?
536
September 13, 2017 08:19AM
Sorry, you can't reply to this topic. It has been closed.
Content reproduced on this site is the property of the respective copyright holders.
It is not reviewed in advance by Oracle and does not necessarily represent the opinion
of Oracle or any other party.