slow query log settings turns on and off constantly

We have a three node cluster. I know we have a query that is much slower than others, I just don't know which one.

With other 8000 queries per second it is also difficult to see anything in real time.

So I want to use the log_slow_queries option. BUT, it keeps flipping between on and off every second...

Second 1:

log_slow_queries..... ON

Second 2:

log_slow_queries..... OFF

Second 3:

log_slow_queries..... ON

Second 4:

log_slow_queries..... OFF

And so on. Everytime this happens the log file on the disk get deleted and recreated too.

In the config files it is turned on. On each of the nodes in the cluster. I never seen anything like this and can only assume it has something to do with the clustercontrol.

Anybody has any ideas? I need to find out what the slow query is but cannot.

Andy

Johan

February 26, 2015 13:06

Hi Andreas.

Yes ClusterControl turn on/off the slow query log. The query is enabled for a period of time (configurable) and the stopped, parsed, erased, recreated .. and so on..

We are considering to change this to allow the slow query log to grow and then read from an offset instead, but now it is like this. And the main reason is to be able to parse and analyze the slow query log in a timely fashion.

Best regards

February 26, 2015 15:11

I should also say that you can of course disable the Query Monitor and deal with the slow query log as you wish:

UI -> Settings -> Query Monitor -> set "Query Sample Time" = -1

Thanks
Johan

Andreas Behr

February 26, 2015 15:45

Ok. I understand now how that works. I was just baffled by that setting always flipping back and forth.

Thanks. I will turn it off in the settings.

slow query log settings turns on and off constantly

Comments

Didn't find what you were looking for?