I ran into a problem. I converted my latin1 table into utf8 with utf8_unicode_ci collation as described in my previous post. The table in question has a UNIQUE index on the utf8_unicode_ci collation column. When reimporting the data I get a duplicate entry on accent e with e itself. Why?

e == utf8 0x65
accented e == UTF-8 0xC3 0xA9, U+00E9

but the rules defined
http://www.unicode.org/reports/tr10/

say to ignore accents for unicode collations. To get around this I know to define my tables as

utf8 with collation of utf8_bin


This bug at mysql.com Here was the indicator that accent e is not the problem.

Stephen O'Grady apparently spent last week at the wrong conference. He came away from OSCON wondering if open source companies can ever reach the $1 billion mark in sales. He writes:

As stated at OSCON, I believe both that open source has reached and in fact passed the tipping point and is here to stay, and that it has a bright future. At this point, however, I do not expect any of the major pure open source firms to challenge the billion dollar threshold any time soon. The numbers - those that are available, at least - simply don't support such a conclusion. But neither do I consider that a metric for success.What's interesting to me is that the very same numbers - at least, I assume we have access to the same data - tell me a very different story. Red Hat will hit the $1B mark in 3-5 years, and will likely do so without seriously tapping into …

Just as a reminder that the MySQL meetup will be happening next Monday at the Elysian.

Directions to the Elysian can be found at:
http://www.elysianbrewing.com/Elysian.html

Our normal range of topics will be covered... aka we talk about MySQL for about half the evening, and then we spend the rest of the evening geeking about other Lamp projects and who has tricked out their MythTV in the last month :)

I'll do a wrap up of what I saw at OSCON last week.

This is a follow-on to Does your organization need a DBA? and Andy Leonard's Database Professionals: An Enterprise Requirement.

I was doing my morning reading and came across this nugget from Jose Barreto:

Maintaining your MOSS 2007 SQL Databases

Mr. Barreto's point is that while Microsoft Office Sharepoint Server (MOSS ) 2007 will do the database creation for you when installed, there still needs to be database maintenance performed. In the case of MOSS 2007, he says there may be as many as 8 databases if …

Parallel with last years international PHP conference the folks at S&S organized a novelty: an open source database conference. It was a great success. We had people from most projects for end users to talk to: PostgreSQL, BerkleyDB, MySQL, SQLite, Apache Derby, HSQLDB, Firebird and MaxDB. I hope this year Ingres will also make it and maybe a few others I do not even know yet! For the projects themselves it was a great opportunity to sit down and discuss about the potential for cooperation. The conference will be held again, same time same place. So mark your calendar November 6th-8th opendbcon 2006 in Frankfurt germany.

The …

I’ve read at times people complaining about the response to bugs, and people bag the support of MySQL on the forums at times.

Well today I logged a bug, not the first and I’m sure it’s not the last. See LAST_INSERT_ID() does not return results for a problem in the latest Connector/J 5.0.3 that was released just recently.

Now it took me about 2 hours to log the bug, and probably at least 2 hours of frustration prior to that. The initial frustration 2 hours was unsuccessful debugging of what I was sure was valid code (and it was near midnight last night). The second 2 hours today was testing the problem between two environments, different database versions and different Connector/J versions, and providing a simple reproducable case of said problem.

So the timeline of the …

I’ve just come back from lunch. I’ve managed to eat Chinese food, in China, with chopsticks and not totally embarass myself. Ate some new food, new vegetables and a seemingly different type of seaweed than I have eaten before. It tasted good though. I even think Kit would have liked some of it (once she got over the fact that it looked different and some things were green things).
I arrived safely after a flight that was fine (except for getting up rather early to get to Sydney to then take a sane timed flight). Beijing seems to be a bit like the firefly world, except with less flying cars. You’ve got heaps of stuff in English and Chinese. It could be really interesting to live here and experience things.

There’s a national English language newspaper which is fairly up to date on world events - the fact that our dear Mr Howard is going to go to the election seems to be news here! It’s not packed with local news, which would …

MySQL is adding more tools to monitor its internals with every new release, but one thing it still lacks is a way to find out who is locking what, and therefore which transactions block which other ones. This is such a vital feature that I'm considering writing my own patch to the source! Still, it is possible, to a limited extent, to find out who's locking resources. In this article I'll explain how you can do that.

This article is the second in a series on how to use the innotop MySQL and InnoDB monitor.

I’ve made several improvements to the innotop InnoDB and MySQL monitor, and it’s ready to get from the innotop homepage. As usual, it’s a combination of bug fixes, new features, and enhancements to existing features. Changes since version 0.1.112 include More detailed lock information displayed in InnoDB Deadlock mode. A new section in Row Operations / Semaphores mode shows information about the wait array. I’ve expanded the test suite a lot.

For any DBA, broken replication represents a dreaded moment. Replication can break for many factors including when the master server issues a 'Duplicate entry' error. At that point you can either skip the slave counter (recommended) or ignore/skip the errors (slave-skip-errors=1062) if you are sure that ignoring such errors won't cause a nasty effect somewhere.

Depending on when you troubleshoot the situation, when replication resumes, slaves may fall significantly behind and the relay log files may start consuming huge amount of disk space. If you have ample free disk space then that is not an issue however, if you are tight on disk space then the relay-log-space-limit option can really come handy.

The relay-log-space-limit option limits the disk space used by the relay log files. You can specify the value in bytes, megabytes (10000M) …