Re: Error=1236 Error reading packet from server for channel 'group_replication_recovery': The slave is connecting using CHANGE MASTER TO MASTER_AUTO_POSITION = 1

[CCGO Programme Office CCGO Programme Office]

Frederic Descamps Wrote:
-------------------------------------------------------
> unfortunately, if one of your nodes was
> disconnected and meanwhile the binary logs with
> the next GTID has been purged, you will have to
> take a backup of an existing member and restore
> it.


Please advice the following procedure is workable

1. stop group_replication for 2nd node (it is the primary server) and set global transaction_write_set_extraction=OFF.
2. At 2nd node, type
mysqldump --all-databases --set-gtid-purged=ON --single-transaction -uroot -S -p > /mysql_data/data/alldb.sql
3. stop group_replication for 1st node (it is the resilience server) and 3rd node (it is the secondary server)
4. Copy the alldb.sql to 1st node and 3rd node
5. At 1st and 3rd node, type
reset master;
source /mysql_data/data/alldb.sql
6. At 2nd node, type
set global transaction_write_set_extraction=XXHASH64;
SET GLOBAL group_replication_bootstrap_group=ON;
START GROUP_REPLICATION;
SET GLOBAL group_replication_bootstrap_group=OFF;
7. At 1st and 3rd node, type
START GROUP_REPLICATION;