… about 180sec first time and about 260sec second time (with innodb_flush_log_at_trx_commit=0) go giving 550 and 380 updates/sec appropriately… you have a lot of tables: 091118 18:43:36 InnoDB: Database was not shut down normally! InnoDB: Starting crash recovery. InnoDB: Reading tablespace…
Post: Innodb usability and ease of use.
September 28, 2007 By 14 Comments
…. 5 years ago Heikki was along and he had a lot to do to make things work well so a… less understandable with value 0 meaning disabling Innodb internal thread queuing. Another one is innodb_flush_logs_at_trx_commit – initially it had only values 0… and 1. Interesting enough in some cases Innodb team does get things right. innodb_flush_method variable does not use value 0,1…
Post: MySQL Crash Recovery
July 30, 2006 By 30 Comments
… found new Innodb bug (1%) Innodb recovery should be automatic and bring your database to consistent state. Depending on innodb_flush_lot_at_trx_commit setting you may lose few last committed transactions…
Post: XtraDB/InnoDB CPU bound benchmarks on 24cores server
December 29, 2008 By 6 Comments
… innodb_buffer_pool_size=15G innodb_data_file_path=ibdata1:10M:autoextend innodb_file_per_table=1 innodb_flush_log_at_trx_commit=2 innodb_log_buffer_size=8M innodb_log_files_in_group=2 innodb_log_file_size=512M innodb_status_file=0 innodb_thread… come. Also in real production there may be not a lot database fits “in-memory” cases – but on other hand 64…
Comment: Choosing proper innodb_log_file_size
July 30, 2006 By 12 Comments
… found new Innodb bug (1%) Innodb recovery should be automatic and bring your database to consistent state. Depending on innodb_flush_lot_at_trx_commit setting you may lose few last committed transactions…
Post: Analyzing the distribution of InnoDB log file writes
July 16, 2010 By 5 Comments
… workloads.) Now, what does this actually mean? There are a lot of interesting and complex things to think about here and… that the writes are smaller than 4k. This server has innodb_flush_log_at_trx_commit set to 2. This means each transaction does a… this up with some more benchmarks or observations under different InnoDB settings and different workloads. Or then again, maybe Yasufumi will…
Post: A case for MariaDB's Hash Joins
May 31, 2012 By Leave a Comment
… Configuration innodb_file_per_table=1 innodb_file_format=barracuda innodb_log_file_size=512M innodb_log_files_in_group=2 innodb_flush_log_at_trx_commit=2 innodb_flush_method… of BKA increases, as there are going to be a lot more random index lookups needed to be performed because of…
Post: SSD, XFS, LVM, fsync, write cache, barrier and lost transactions
March 2, 2009 By 32 Comments
… innodb_log_file_size=256M innodb_thread_concurrency=0 innodb_flush_log_at_trx_commit=1 innodb_flush_method = O_DIRECT Actually most interesting one are innodb_flush_log_at_trx_commit=1 and innodb_flush… LVM is putting barrier option which decreases write performance a lot
Post: MySQL 5.5.8 and Percona Server: being adaptive
December 20, 2010 By 25 Comments
…innodb_buffer_pool_size=24G, innodb_log_file_size=2000M (innodb_log_files_in_group=2), and innodb_flush_log_at_trx_commit=2. Also, innodb_adaptive_flushing (ON) / innodb…
Post: How much overhead DRDB could cause ?
June 2, 2008 By 4 Comments
… why bother with DRBD at all and not use MySQL replication instead. So assuming you have innodb_flush_log_at_trx_commit=1 and sync_binlog… record in Innodb log to commit it and finally there is record in Binary log to commit transaction followed by commit transaction in Innodb storage… point is there is a lot of synchronous writes which are fully serialized because group commit in Innodb was broken in 5.0…
