Percona XtraDB and crash-safe replication

Hi,

Is it possible to enable crash-safe replication on Percona XtraDB cluster?

When I tried to add the following variables the MySQL server refuses to start and it constantly print the same message in the error log:

/etc/my.cnf

relay_log_info_repository = TABLE
relay_log_recovery = 1

Error log:

It works when i tested this with standard Percona server
OS: CentOS 6.5
MySQL : Percona-XtraDB-Cluster-server-56-5.6.15-25.5.759

What is the purpose you are trying to enable it? Is any of the cluster members also an asyc slave of another mysql instance?

I’m not using Galera feature for now, just standard MySQL replication hence I would like to enable this option.

I cannot reproduce this kind of crash, can you post your full my.cnf? Does it crash right after mysqld start or only once some some trx is applied via replication?

It crashes, right after mysqld start.

[mysqld_safe]
log-error = /var/log/mysql/mysql-error.log

[mysqld]
user = mysql
port = 3306
bind_address = 0.0.0.0
socket = /var/lib/mysql/mysql.sock

datadir = /var/lib/mysql
tmpdir = /tmp
pid-file = /var/lib/mysql/mysql.pid

character-set-server = utf8
explicit_defaults_for_timestamp = 1

slow-query-log = 1
slow_query_log_file = /var/log/mysql/mysql-slow.log
long_query_time = 5

log-bin = /var/lib/mysql/log/mysql-bin
binlog-format = MIXED
expire_logs_days = 14
relay-log = /var/lib/mysql/log/mysql-relay
relay_log_info_file = /var/lib/mysql/log/relay-log.info
slave_load_tmpdir = /var/lib/mysql/log/tmp
server-id = 737731821

# Crash-safe replication
master_info_repository = TABLE
relay_log_info_repository = TABLE
relay_log_recovery = 1

#skip-slave-start

read_only

back_log = 50
max_connections = 500
max_connect_errors = 100
skip-name-resolve

key_buffer_size = 256M
sort_buffer_size = 2M
read_rnd_buffer_size = 4M
bulk_insert_buffer_size = 128M
max_allowed_packet = 32M
max_heap_table_size = 64M
tmp_table_size = 64M

table_open_cache = 2048
thread_cache_size = 16

query_cache_type = 0
query_cache_size = 0

# InnoDB
innodb_data_home_dir = /var/lib/mysql
innodb_log_group_home_dir = /var/lib/mysql
innodb_data_file_path = ibdata1:2000M:autoextend
innodb_flush_log_at_trx_commit = 2
innodb_flush_method = O_DIRECT
innodb_log_buffer_size = 16M
innodb_buffer_pool_size = 512M
innodb_log_file_size = 512M
innodb_log_files_in_group = 2
innodb_read_io_threads = 8
innodb_write_io_threads = 8
innodb_io_capacity = 400
innodb_max_dirty_pages_pct = 90
innodb_lock_wait_timeout = 120

[mysql]
no-auto-rehash
socket=/var/lib/mysql/mysql.sock

[client]
socket=/var/lib/mysql/mysql.sock

OK, binlog-format = MIXED was the key here. Please change it to ROW (as MIXED is not supported by Galera anyway), and try enabling crash safe again.

Bug reported here: [url]https://bugs.launchpad.net/percona-xtradb-cluster/+bug/1313756[/url]