Thank you for the advice, and I read about "mysqlbinlog"....
>>>> 2011/08/15 10:22 -0400, Shawn Green (MySQL) >>>>
1) extract the statements from the binary log.
2) get the master to sent you the statements just as if you were a slave.
mysqlbinlog will do either - (again, read the manual on how to use the tool)
Once you have collected the statements you need the slave to apply (and put them on your
flash drive), now it's your turn to replay those statements on the slave. The easiest tool
for that will probably be the mysql client (a command-line tool). This is where you become
the SLAVE SQL thread.
Beyond that, all you really need to keep up with is the binary log position you replicated
last (again, pretending to be the SLAVE IO thread).
Best of luck! what you are doing is definitely labor intensive.
I do not understand 1) or 2). Playing with "mysqlbinlog" I see that I can get earlier or
later bunches of SQL commands if not all of them, but it seems to me that it is not safe
to use part of its output unless carefully done, for, say, I use ANSI mode and its output
uses the character ere the small "a" for enclosing variable names, and local "sql_mode" is
changed. I also started "mysqld" with the "log-bin" argument something convenient for me
I suppose you mean that for finishing the replication I make the "mysqlbinlog" output
"mysql" s standard input.
I find a later starting point by noting the greatest "Xid" value?
In this case it is not very labor-intenstive, for the transaction rate on these databases
is most readily noted in inverse weeks, or maybe inverse longer whiles.