Showing results for 
Show  only  | Search instead for 
Did you mean: 

This product reached the end of support date on March 31, 2021.

IBM WAS Wait time issue


As the image. Why wait time long?

How can I find the root-cause?



DynaMight Leader
DynaMight Leader

Hello Alex,

Below is the formula to calculate the response time.

Response Time Formulas

When a PurePath is executed, AppMon captures CPU time used by the threads that are executing this transaction. It also captures the information on how much time was spent in sync (sync blocks), wait (waiting on an object) and runtime suspension (when garbage collection pauses the thread). The rest of the time is contributed by I/O.

The formula for response time therefore is:

Response Time = CPU + Sync + Wait + Runtime Suspension + I/O

I/O is calculated and everything else is measured:

I/O = Response Time - CPU - Sync - Wait - Runtime Suspension

Right click on your PurePath and Drill Down the Response Time Hotspots to check which API and Tier is involved and then check the API Breakdown and Method Hotspots from the same dashlet.

In your case it is waiting for the object.



Dynatrace Pro
Dynatrace Pro

Hi Alex,

from what I can see in your purepath, the wait time is a nice round value of exactly 60 seconds which looks like a time out. It is making some kind of external call to another system (third party or JVM with no agent) and my guess is that it is not responding in time.

There is a log SEVERE statement below, and I can only see the beginning of the message but it looks like what has been returned cannot be parsed. Typically it would be a server down or time out message which is not what the code is expecting and therefore fails while parsing it.

Calls via HTTP to external systems should be automatically detected but in your case it doesn't seem to be the case unless its higher up in your purepath. Could you attach the whole purepath in dts format and this way I could check it for you?



Adding to what Florent said, waiting for the other end to take the request and start to process could be related to unavailable threads. Calls via HTTP is treated by Web Container threads. If you have all tiers of an application monitored you could create a PMI measure (need to be enabled at WAS) to show the number of used threads by JVM and see if at the moment of a problem the threads available were exhausted.