0

Background:

I'm working on dataview, and many of the reports are generated by very long running queries. I've written a small query caching daemon in python that accepts a query, spawns a thread to run it, and stores the result when done as a pickled string. The results are generally various aggregations broken down by month, or other factors, and the result sets are consequently not large. So my caching daemon can check whether it has the result already, and return it immediately, otherwise it sends back a 'pending' message (or 'error' or 'failed' or various other messages). The point being, that the client, which is a django web server would get back 'pending' and query again in 5~10 seconds, in the meanwhile putting up a message for the user saying 'your report is being built, please be patient'.

The problem:

I would like to add the ability for the user to cancel a long running query, assuming it hasn't been cached already. I know I can kill a query thread in MySQL using KILL, but is there a way to get the thread/query/process id of the query in a manner similar to getting the id of the last inserted row? I'm doing this through the python MySQLdb module, and I can't see any properties/methods of the cursor object that would return this.

4

1 に答える 1

2

There is a property of the connection object called thread_id, which returns an id to be passed to KILL. MySQL has a thread for each connection, not for each cursor, so you are not killing queries, but are instead killing connection. To kill an individual query you must run each query in it's own connection, and then kill the connection using the result from thread_id

于 2012-10-05T09:32:06.363 に答える