Threads not stopped when client stopped

[Terry Gliedt]

I'm posting here because the database in question is a MyISAM database, but I suspect the problem is broader than this.. We are testing for massively large genetic databases (> 700M rows, about 80GB in size) and are using MySQL 4.1.11-Debian_4sarge2.

When someone begins a query which might take hundreds of seconds to complete, often they will control-C and cancel the application. I've recently discovered that the thread that has been initiated by the query continues to run on the server. If it only had a small consequence, I would not care, but these queries often result in many GB of temp space being used.

Failing to stop the thread certainly seems like incorrect behavior to me. Before I report this as a bug, has anyone else seen this? Is there anything I can do about this right now? It's killing my server fetching data which all too often is not even received by the client. TIA