my.cnf configuration from percona xtraDB cluster 5.7 to percona xtraDB cluster 8.0.19

Hi,
We are going to migration from percona xtraDB cluster 5.7 to percona xtraDB cluster 8.0.19, they both on different box (as we change OS, so must be on other host) so we are not going to do do in place upgrade.

in the existing my.cnf of percona xtraDB cluster 5.7 (see below) we have this global variable, not sure which one is not needed anymore and what should be the optimized global variable for us on percona 8.0.19:

[mysqld]#general_log = on #general_log_file=/var/log/mysql/query#datadir=/mysql/datadatadir=/var/lib/mysqlsocket=/var/lib/mysql/mysql.sockuser=mysql  (is it still needed?)
# Path to Galera librarywsrep_provider=/usr/lib64/galera3/libgalera_smm.so
# Cluster connection URL contains the IPs of node#1, node#2 and node#3wsrep_cluster_address=gcomm://<node#1 IP>,<node#2 IP>
# In order for Galera to work correctly binlog format should be ROWbinlog_format=ROW
# MyISAM storage engine has only experimental supportdefault_storage_engine=InnoDB
# This InnoDB autoincrement locking mode is a requirement for Galerainnodb_autoinc_lock_mode=2
# Node 1 addresswsrep_node_address=<node#1 IP>
# SST methodwsrep_sst_method=xtrabackup-v2
# Cluster namewsrep_cluster_name=<cluster name>
#Authentication for SST methodwsrep_sst_auth=“user:password”
# MySQL VARIABLESmax_connections = 10000key_buffer_size =256000000myisam_sort_buffer_size=67108864read_buffer_size =1048576read_rnd_buffer_size=4194304sort_buffer_size=1048576query_cache_size=16777216
lock_wait_timeout=300wait_timeout=28800interactive_timeout=28800connect_timeout=30
sort_buffer_size=256000000max_allowed_packet = 1073741824
lower_case_table_names=1autocommit=0open_files_limit=10000
log_timestamps = SYSTEM

transaction_isolation="READ-COMMITTED"sql_mode=“ONLY_FULL_GROUP_BY,STRICT_TRANS_TABLES,NO_ZERO_IN_DATE,ERROR_FOR_DIVISION_BY_ZERO,NO_AUTO_CREATE_USER,NO_ENGINE_SUBSTITUTION”


I would suggest you read up on the configuration differences between 5.7 and 8.0. The MySQL 8.0 manual has much detail here. It is not safe to simply copy 5.7 my.cnf over to 8.0. Many variables were removed, and some 5.7 defaults are no longer valid in 8.0.

" It is not safe to simply copy 5.7 my.cnf over to 8.0. Many variables were removed, and some 5.7 defaults are no longer valid in 8.0."
this is what I am wondering, I tried to put old setting to new cluster, then I must bootstrap it again, right?

You only have to bootstrap the first node of the cluster. If your cluster is currently down, yes, edit the config, remove any old 5.7 variables, then bootstrap the node.

so after adding old 5.7 variables, bootstrap the first node and test.  
if doesn’t work, remove the old 5.7 variables, then bootstrap the first node again and store back to normal ?

"he configuration differences between 5.7 and 8.0. "
any resource for me to read ?

Check the MySQL manual for upgrading from 5.7 -> 8.0. The manual lists every parameter that was removed.