tag:blogger.com,1999:blog-768233104244702633.post2348884547821051952..comments2023-11-16T03:16:54.746-08:00Comments on The Scale-Out Blog: The Agony of Big Transactions in the MySQL BinlogRobert Hodgeshttp://www.blogger.com/profile/05379726998057344092noreply@blogger.comBlogger3125tag:blogger.com,1999:blog-768233104244702633.post-27328615290188595112012-02-11T21:51:05.213-08:002012-02-11T21:51:05.213-08:00I saw this when using huge LOAD DATA INFILE import...I saw this when using huge LOAD DATA INFILE imports.. all other commits are blocked while the binlog gets written.sbesternoreply@blogger.comtag:blogger.com,1999:blog-768233104244702633.post-58019826104701306952012-02-06T11:21:12.668-08:002012-02-06T11:21:12.668-08:00Hi Sheeri! The workload is largely statements with...Hi Sheeri! The workload is largely statements with some RBR to avoid repeating large transactions on slaves. As far as I know the large transaction in question consisted of many individual statements, which allows Tungsten to fragment. The really enormous transactions I have seen (like the 5 GB example) were however RBR.Robert Hodgeshttps://www.blogger.com/profile/05379726998057344092noreply@blogger.comtag:blogger.com,1999:blog-768233104244702633.post-71060455813747192012-02-06T10:57:28.451-08:002012-02-06T10:57:28.451-08:00It seems like this was a statement-based log, beca...It seems like this was a statement-based log, because you took out something and replaced it with (SQL Statement) but was the SQL statement really over 1 Gb?<br /><br />Or was this row-based binary logging/replication?Sheeri K. Cabralhttps://www.blogger.com/profile/13990877688502800403noreply@blogger.com