Comments on: Upgrading to MySQL 5.1 GA? Better Wait http://thinkapi.com/blog/upgrading-to-mysql-51-ga-better-wait/ An interface to tech thoughts Sat, 17 Dec 2011 07:37:55 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: Sukumar http://thinkapi.com/blog/upgrading-to-mysql-51-ga-better-wait/comment-page-1/#comment-6 Sukumar Mon, 01 Dec 2008 18:27:15 +0000 http://thinkapi.com/blog/?p=89#comment-6 @Sheeri Thanks a lot for your views. For your last question, the last link in my article is perhaps the answer :) http://www.mysqlperformanceblog.com/2008/06/30/mysql-51-stability/ @Sheeri

Thanks a lot for your views. For your last question, the last link in my article is perhaps the answer :)
http://www.mysqlperformanceblog.com/2008/06/30/mysql-51-stability/

]]> By: Sheeri Cabral http://thinkapi.com/blog/upgrading-to-mysql-51-ga-better-wait/comment-page-1/#comment-5 Sheeri Cabral Mon, 01 Dec 2008 15:48:05 +0000 http://thinkapi.com/blog/?p=89#comment-5 Monty is responsible for Maria, which isn't in 5.1. Monty's points are all valid, but they're nothing new. They're just saying what every DBA and sysadmin knows -- that the first GA of a release isn't perfect. It was like that with the first release of 4.0, and 4.1, and 5.0. Also, the point about testing it for a few weeks -- *no* upgrade should be done without adequate testing! I can tell you that we have clients using 5.1 in production, and have been for a few months. They tested, found it to suit their needs, and use it, without random unexpected crashes (we did have one 5.1 server crash when we were doing lots of DW stuff and used too much memory, but that's our own fault, and would have happened in 5.0). While I agree with all the points made, I don't see them as anything other than "I need to do my job responsibly as a DBA, to make sure that this release will work in this environment." That includes looking at the release notes in the manual to see what's changed, to see if it's worth an upgrade, and testing, to make sure there isn't some odd bug that just happens to be snagged by this particular environment. Zero critical bugs *is* impossible, for just about any complex software. It's not a WTF that it's mentioned, but perhaps it's a bad PR move. Honesty and transparency is a good thing though. I'm just wondering though: Which production systems would you install an upgrade on without actually testing for a few weeks? Monty is responsible for Maria, which isn’t in 5.1. Monty’s points are all valid, but they’re nothing new. They’re just saying what every DBA and sysadmin knows — that the first GA of a release isn’t perfect.

It was like that with the first release of 4.0, and 4.1, and 5.0.

Also, the point about testing it for a few weeks — *no* upgrade should be done without adequate testing!

I can tell you that we have clients using 5.1 in production, and have been for a few months. They tested, found it to suit their needs, and use it, without random unexpected crashes (we did have one 5.1 server crash when we were doing lots of DW stuff and used too much memory, but that’s our own fault, and would have happened in 5.0).

While I agree with all the points made, I don’t see them as anything other than “I need to do my job responsibly as a DBA, to make sure that this release will work in this environment.” That includes looking at the release notes in the manual to see what’s changed, to see if it’s worth an upgrade, and testing, to make sure there isn’t some odd bug that just happens to be snagged by this particular environment.

Zero critical bugs *is* impossible, for just about any complex software. It’s not a WTF that it’s mentioned, but perhaps it’s a bad PR move. Honesty and transparency is a good thing though.

I’m just wondering though: Which production systems would you install an upgrade on without actually testing for a few weeks?

]]>