Re: (mysqldump) Serial output. . .?

2006-02-21 Thread SGreen
Michael, I have been following this thread from the beginning and I just don't see the practical difference between what you propose and the replication methods (SBR and RBR) already in place. How does what you propose differ from the SBR (statement -based replication) that MySQL already

Re: (mysqldump) Serial output. . .?

2006-02-21 Thread mwilliams
Shawn, Thanks for the reply, but I think the thread has become much more dramatic than nececessary. Basically, I'm not looking for what has been ALTERed. I simply need the table creation data output in ALTER IGNORE (or whatever is appropriate) format so as to ensure table structure is the

Re: (mysqldump) Serial output. . .?

2006-02-21 Thread SGreen
One problem with dual-master or multi-master replication is that you have to be able to set and check a lock across all masters before performing a schema change. How would you deal with this scenario using your ALTER TABLE database dumps without such a lock? Server A and B share a table X

Re: (mysqldump) Serial output. . .?

2006-02-21 Thread mwilliams
Ah, well, in this particular DB, *every single transaction* it's it's own entry. . .basically the DB itself is a binary log. . .kinda. . .sorta. So the current value of a particular item isn't necessarily an issue since, once entered, it will always be the same. A change to that value will in

Re: (mysqldump) Serial output. . .?

2006-02-21 Thread SGreen
Ok, I get that. I have several tables just like that (I use mine as shadow tables for change audits. Every change to the normal table ends up creating new record in the shadow table thus documenting each state of the normal table through time). However, shouldn't schema changes be very rare

Re: (mysqldump) Serial output. . .?

2006-02-21 Thread mwilliams
Yes, schema changes would/should be rare. You may be right, but I would think that since schema information is *alway* checked before any INSERTS then we should be good. There should never be an occasion to break per se. Obviously anything can happen, and appropriate recovery methods (e.g.

Re: (mysqldump) Serial output. . .?

2006-02-20 Thread sheeri kritzer
mysqldump takes a table or database and dumps it -- current schema, current data. You won't get alter tables. What you want is something that will show all the alter statements. You can run something like this on unix: tail -f binlog* | grep ALTER alter.sql and then the alter.sql text file

Re: (mysqldump) Serial output. . .?

2006-02-20 Thread mwilliams
Sheeri, Thanks very much for the reply. However, that is not what I'm looking for. I don't want the tables that *have been* altered. I want CREATE TABLE statements output in the ALTER TABLE format for re-creation on another system (clean or otherwise). Regards, Michael -- Original

RE: (mysqldump) Serial output. . .?

2006-02-17 Thread mwilliams
PROTECTED] Cc: mysql@lists.mysql.com Subject: RE: (mysqldump) Serial output. . .? Shawn, I'd actually considered such, but I was hoping for it to already be present. I'm currently on a tight deadline to finish a project I'm working on so devoting time to getting 'mysqldump' stable

RE: (mysqldump) Serial output. . .?

2006-02-16 Thread Dathan V. Pattishall
Mysqldump can do all the following except dump the ALTER commands, it will log the create statement of the table as the table exists. Mysql doesn't keep a record of what, when, how a table was altered only the final result. Look at the mysqldump options for the stuff you want to do by typing

RE: (mysqldump) Serial output. . .?

2006-02-16 Thread mwilliams
Thanks Dathan, But I'm more concerned with ALTER than anything. As I said, I basically want to be able to output field, type, etc. by row so that *if* the receiving DB needs to add a field it can do so without having to nuke the table with all data in it just to add all fields again with a

RE: (mysqldump) Serial output. . .?

2006-02-16 Thread SGreen
My suggestion: Modify the source of mysqldump yourself. After all, it is open source. Make sure you adhere to any and all licensing requirements and copyright notices and you will keep yourself out of any legal trouble. For the vast majority of users, replication is a better solution than

RE: (mysqldump) Serial output. . .?

2006-02-16 Thread mwilliams
Shawn, I'd actually considered such, but I was hoping for it to already be present. I'm currently on a tight deadline to finish a project I'm working on so devoting time to getting 'mysqldump' stable enough to then propagate across corporate servers in such a short period is not very

RE: (mysqldump) Serial output. . .?

2006-02-16 Thread Logan, David (SST - Adelaide)
@lists.mysql.com Subject: RE: (mysqldump) Serial output. . .? Shawn, I'd actually considered such, but I was hoping for it to already be present. I'm currently on a tight deadline to finish a project I'm working on so devoting time to getting 'mysqldump' stable enough to then propagate across corporate

RE: (mysqldump) Serial output. . .?

2006-02-16 Thread Dirk Bremer
- From: mwilliams [mailto:[EMAIL PROTECTED] Sent: Thursday, February 16, 2006 14:44 To: [EMAIL PROTECTED] Cc: mysql@lists.mysql.com Subject: RE: (mysqldump) Serial output. . .? Shawn, I'd actually considered such, but I was hoping for it to already be present. I'm currently