[MDEV-5146] Bulk loads into partitioned table not working. Created: 2013-10-16  Updated: 2015-08-21  Resolved: 2015-08-21

Status: Closed
Project: MariaDB Server
Component/s: wsrep
Affects Version/s: 5.5.33a-galera
Fix Version/s: 5.5.46-galera, 10.0.22-galera

Type: Bug Priority: Major
Reporter: Matthew Wheeler Assignee: Nirbhay Choubey (Inactive)
Resolution: Fixed Votes: 2
Labels: None
Environment:

Centos 6.3 64 bit.



 Description   

We have:
wsrep_load_data_splitting=on
in our server.cnf file. Bulk loads into a partition table do not get replicated to the other nodes.

Log file shows:

131007 10:38:54 [Warning] WSREP: SQL statement was ineffective: LOAD DATA INFILE '/workspace3/tempspace/OnTargetPublish.load_file.24194.20131007103841' IGNORE
 
INTO TABLE TRIGGERED_SENDS.LOG_BASED_SENDS_UNIV
 
FIELDS TERMINATED BY '  '
 
LINES TERMINATED BY '\n'
 
(
 
        EMAIL_KEY,
 
        SEND_ID,
 
        @START_DTTM,
 
        @END_DTTM,
 
        ACTIVE_IND,
 
        CHR_DATA_OBJ
 
)
 
SET START_DTTM = STR_TO_DATE(@START_DTTM,'%Y%m%d%H%i%s'),
 
END_DTTM = STR_TO_DATE(@END_DTTM,'%Y%m%d%H%i%s'),
 
EVENT_DTTM = NULL,
 
TO_DB_DTTM = NOW(),
 
UPDT_DB_DTTM = NOW()
 
 => Skipping replication

I think that may be normal - we want the records replicated, but not the sql statement itself.

The table is partitoned on SEND_ID.
I'm not sure if it is the partition table that is causing the issue or something else.

Let me know if you need anything else to debug or have anything else you want us to try.
Thanks.

 Comments   
Comment by Elena Stepanova [ 2013-10-16 ]

Hi Matthew,

Could you please provide the full cnf file(s)?
Thanks.

Comment by Matthew Wheeler [ 2013-10-16 ]

from the node that it is submitted to - 4c-maria-01:

 
[server]
 
 
 
[mysqld]
 
socket          = /database/mysql.sock
 
port            = 3306
 
datadir         = /database/data
 
tmpdir          = /database/tmp/
 
innodb_log_group_home_dir = /database/logFiles
 
innodb_data_home_dir = /database/data
 
 
 
# logs and replication
 
log-bin=mysql-bin
 
binlog-format=ROW
 
 
 
server-id=175
 
 
 
max_binlog_cache_size=1024G
 
 
 
max_connections=255
 
max_connect_errors=10000
 
max_allowed_packet=128M
 
 
 
# Galera Settings
 
wsrep_provider=/usr/lib64/galera/libgalera_smm.so
 
wsrep_cluster_address=gcomm://10.200.0.11
 
wsrep_cluster_name='client01_cluster'
 
wsrep_node_name='4c-maria-01'
 
wsrep_slave_threads=24
 
wsrep_retry_autocommit=10
 
wsrep_sst_method=xtrabackup
 
wsrep_sst_auth=galera:somepassword
 
wsrep_load_data_splitting=on
 
#wsrep_notify_cmd=/bin/wsrep_notify.sh
 
 
 
# innodb (xtradb) settings
 
default_storage_engine=InnoDB
 
innodb_file_per_table
 
innodb_file_format=barracuda
 
innodb_log_file_size=2000M
 
innodb_log_files_in_group=2
 
innodb_flush_log_at_trx_commit=2
 
innodb_autoinc_lock_mode=2
 
innodb_locks_unsafe_for_binlog=1
 
 
 
# For zfs we cannot use O_DIRECT
 
#innodb_flush_method=O_DIRECT
 
 
 
# For zfs native aio fails on tmpfs
 
innodb_use_native_aio=FALSE
 
 
 
# zfs prevents partial writes already.  No need to have MariaDB issue them.
 
innodb_doublewrite=0
 
 
 
innodb_io_capacity=1400
 
innodb_thread_concurrency=0
 
innodb_read_io_threads=24
 
innodb_write_io_threads=24
 
innodb_purge_threads=1
 
 
 
innodb_buffer_pool_size=44G
 
innodb_buffer_pool_instances=8
 
innodb_additional_mem_pool_size=20M
 
 
 
[mariadb-5.5]
 
thread_handling=pool-of-threads

and the second node - 4c-maria-02:

 
[server]
 
 
 
[mysqld]
 
socket          = /database/mysql.sock
 
port            = 3306
 
datadir         = /database/data
 
tmpdir          = /database/tmp/
 
innodb_log_group_home_dir = /database/logFiles
 
innodb_data_home_dir = /database/data
 
 
 
# logs and replication
 
log-bin=mysql-bin
 
binlog-format=ROW
 
 
 
server-id=176
 
 
 
max_binlog_cache_size=1024G
 
 
 
max_connections=255
 
max_connect_errors=10000
 
max_allowed_packet=128M
 
 
 
# Galera Settings
 
wsrep_provider=/usr/lib64/galera/libgalera_smm.so
 
wsrep_cluster_address=gcomm://10.200.0.10
 
wsrep_cluster_name='client01_cluster'
 
wsrep_node_name='4c-maria-02'
 
wsrep_slave_threads=24
 
wsrep_retry_autocommit=10
 
wsrep_sst_method=xtrabackup
 
wsrep_sst_auth=galera:somepassword
 
wsrep_load_data_splitting=on
 
#wsrep_notify_cmd=/bin/wsrep_notify.sh
 
# Slave Settings
 
wsrep_provider_options="gcs.fc_limit=256;gcs.fc_factor=0.9;gcs.fc_master_slave=yes"
 
 
 
# innodb (xtradb) settings
 
default_storage_engine=InnoDB
 
innodb_file_per_table
 
innodb_file_format=barracuda
 
innodb_log_file_size=2000M
 
innodb_log_files_in_group=2
 
innodb_flush_log_at_trx_commit=2
 
innodb_autoinc_lock_mode=2
 
innodb_locks_unsafe_for_binlog=1
 
 
 
# For zfs we cannot use O_DIRECT
 
#innodb_flush_method=O_DIRECT
 
 
 
# For zfs native aio fails on tmpfs
 
innodb_use_native_aio=FALSE
 
 
 
# zfs prevents partial writes already.  No need to have MariaDB issue them.
 
innodb_doublewrite=0
 
 
 
innodb_io_capacity=1400
 
innodb_thread_concurrency=0
 
innodb_read_io_threads=24
 
innodb_write_io_threads=24
 
innodb_purge_threads=1
 
 
 
innodb_buffer_pool_size=44G
 
innodb_buffer_pool_instances=8
 
innodb_additional_mem_pool_size=20M
 
 
 
[mariadb-5.5]
 
thread_handling=pool-of-threads

Comment by Nirbhay Choubey (Inactive) [ 2015-08-21 ]

http://lists.askmonty.org/pipermail/commits/2015-August/008271.html

Comment by Jan Lindström (Inactive) [ 2015-08-21 ]

ok, to push.

Comment by Nirbhay Choubey (Inactive) [ 2015-08-21 ]

https://github.com/MariaDB/server/commit/4ee28865f67c980848bb62f0009440be73ebee7c

Generated at Thu Feb 08 07:02:02 UTC 2024 using Jira 8.20.16#820016-sha1:9d11dbea5f4be3d4cc21f03a88dd11d8c8687422.