If you’re interested, you can examine the remarkably short release notes for the bug fix. In a sense, it’s the brevity of the release notes that has made some unhappy. I’ll get to that in a moment; first, let’s discuss the nature of the bug itself.
This particular bug was cited late last month by Hartmut Holzgraefe. “An UPDATE that should only modify a single row runs forever and blows up table / tablespace size until there is no more disk space available,” he bluntly states. You’ll see this problem when you update tables using InnoDB, which is the storage engine used by default for every version of MySQL since 5.5.
It’s a good thing that Oracle wiped out this bug pretty quickly. It may not be the ugliest bug in the world, but anything that uses up all your hard disk space unnecessarily is pretty nasty in my book. On the other hand, for some people, like Peter Laursen, the bug fix doesn’t quite close the issue. As he notes in his blog, “it is still unclear what exact conditions will trigger the effect.”
Oracle’s release also leaves another question unanswered: is there an easy fix for users affected by the bug? Laursen notes that the only solution he can find for affected users is to “dump everything and restore to a fresh server or InnodDB instance” – not something a database administrator would undertake lightly.
Actually, the situation is not quite that bad. H Online notes that “If the disk space is being used by temporary tables, one possibility is to simply stop the server, empty the Temp directory and then restart the server.” But that’s a specific case, and it won’t work if InnoDB tables are eating up your disk space.
Sadly, what may be worse than the bug is Oracle’s attitude toward it. It’s good that they fixed it quickly, but the company wasn’t transparent enough to satisfy some. “The lack of willingness of Oracle here to expose the exact conditions for this bug to affected users (and how to repair it should they be affected) is the real problem here,” Laursen notes. If you have not yet been affected by the problem, it’s easy enough to get the new version with the patch, but Oracle isn’t exactly forthcoming with recommended procedures if the bug has already eaten your hard disk space.
Has your system experienced problems due to this bug? How did you go about repairing the problem? Do you think Oracle should have provided more details? Feel free to comment below.