At the November MySQL User Group, Patrick Galbraith ran into a problem where binlog-do-db was duplicated. It manifests itself like this: (copied from http://lists.mysql.com/replication/607)

master my.cnf: binlog-do-db=db1

slave my.cnf: replicate-do-db=db1;

Relevant show slave status output

Replicate_Do_DB: db1;,db1;

When db1 is modified on master, Read_Master_Log_Pos and Relay_Log_Pos do changes, also I can open and see the changes in rh3-relay-bin.000002, but they do not appear in [...]

The 79th edition of Log Buffer, the weekly review of database blogs, has been published on H. Tonguç Yilmaz Oracle Blog. Next week, LB#80 will appear on Chen Shapira’s I’m just a simple DBA on a complex production system. Have you done one yet? Publishing Log Buffer is an excellent way to introduce yourself and your [...]

Tips on load testing your web site.

read more

When you're storing text of significant size in the table it often makes sense to keep it compressed. Unfortunately MySQL does not provide compressed BLOB/TEXT columns (I would really love to have COMPRESSED attribute for the BLOB/TEXT columns which would make them transparently compressed) but you well can do it yourself by using COMPRESS/UNCOMPRESS functions or compressing/decompressing things on the client.

This choice of server side vs client size compression depends on the queries you have. If you just retrieve these BLOB/TEXT columns for say displaying it in application compressing on application is typically better idea because you will save CPU on database box and move it to the Web box which is typically easier to scale, plus you save on network traffic a bit, though this is not generally the problem.

Compression on the server gives you ability to do some server side filtering (which however would not be overly …

I'm working on an implementation of a mysql slave precache tool as described by Paul Tuckfield at last year's User conference.

I can easily tell where the SQL Thread is currently is reading from in the local relay log, but it's a harder problem to solve when I want to scan ahead in that log some number of binlog entries and find a valid position. I can simply add to the position some number of bytes, but then I have to stop and loop through calls to 'mysqlbinlog' until I get a position that doesn't throw an error.

This turns out to be a pretty expensive operation, and I'm having a problem where by the time I find a good position and can start my precacher, the slave is already past me!

Does anyone have any clever ways to find good binary log positions besides this (this is in ruby, in case you're wondering):

def …

I was writing up some notes about a client’s status variables, and I’d had to remind myself of Handler_read_first. Handler_read_first is a counter that increments when the first entry in the index is read, which may indicate that a full index scan is being used. With this particular client, I ran a SHOW [...]

One of the major projects I worked on last year was the MySQL 5.1 Cluster DBA certification. I did the background research to come up with a sensible scope and outline of the exam, wrote most of the questions and actively looked for feedback from our cluster engineering and development team as well as consultants that are deploying MySQL Cluster in real-world situation.

For example, I got great feedback on the exam from MySQL Cluster Senior Software Engineer Stewart Smith...Response from the exam is overwhelmingly positive and as it?s new, afterwards I discussed a few questions with Roland (in fact, all the ones I got wrong). The (good!?) news is that I legitimately brain-farted on some.

... this isn?t some trivial …

Fortune magazine has published a list of its hot IPO tips for 2008. Three out of the five - MySQL, Ingres and SugarCRM - are open source companies, while another - Parallels - is an open source project sponsor (for the record, the other Fortune tip is ExactTarget). Here’s a look at Fortune’s assessment of the four open source-related vendors, together with a quick 451 CAOS Theory view, and a terrible pun.

I should probably state for the record that I have no knowledge of …

CMSWire asked me to be the first they interview for their brand new column Flash Quiz. If you ever wondered how much I sleep and whether I like fancy cars, check out Flash Quiz: Sandro Groganz Speaks.

So I finally installed the xdebug PHP extension and started doing some performance analysis of MemberDB using xdebug and kcachegrind. The upshot of which is a number of commits to the bzr tree that dramatically improve performance in several key areas. The answer? Caching.

I’m not even talking using memcached or caching things in database tables or anything like that - just about everything is still the same dynamically produced content as before, but I’m now caching some simple things avoiding many round-trips to the database while executing a script.

There were a few things that were taking a fair bit of execution time:

1. The generation of the menu. In MemberDB, there’s a menu on the left. There’s also a …