I keep getting asked to 'go and find something' within the logs whenever our applications hangs/spikes/etc on the client side.

I know to check the db2diag.log and capture current snapshots, but upper mgmt still wants to know what happened on the transaction level. I tell them that you can't read the transaction logs, but they still persist on trying to figure out what happens.

I understand that I can set up event monitors, but they don't want to do that due to performance concerns. What else can I look at/do to provide the level of information they are looking for?

BTW:

LUW 9.5.3
DIAGLEVEL = 3
NOTIFYLEVEL = 3

The diag log isn't showing the -911 errors that are showing up in the application logs along with other 'db2' errors.

What else can I provide?

Your question is too general. Tools that you use depend on the problem that you are trying to investigate.

SQL0911N could indicate either deadlocks or lock timeouts. For the former you use the deadlock event monitor, which has very little overhead and is created and enabled by default - all you need to do is to read the files. For lock timeouts there is additional monitoring that was introduced in DB2 9.5 - you can read about it here: Lock timeout reporting

You should explain to your bosses that there's no such thing as free lunch. If they want to monitor the system all the time, they will pay the price of performance impact all the time.

You do not want the db2diag.log file to fill up with every application error (such as -911). It would be huge in no time and searching through that would be a nightmare.

If you are getting timeouts, you can user a new feature in V9.5:

IBM DB2 9.5 Information Center for Linux, UNIX, and Windows

Any other error you really need the event monitor.

It has been my experience that error -911 (either deadlock or timeout) are because of either poor programming or networking issues.

Andy

Thanks guys....I'm looking at this right now.

Questions I have are

* What is the regex of the output filename resulting from this registry value being set?
* Does the switch ON result in immediate usage of the capability, or do I need to recycle the instance?

The information provides doesn't address any of this.

Thanks.

Here is some more information.

Lock events for DB2 for Linux, UNIX, and Windows, Part 2: New options for analyzing lock timeouts in DB2 9.5

I cannot remember if you have to recycle the instance. My guess is yes.

Andy