Question
Incessant Technologies Pvt Ltd.
AU
Last activity: 9 Feb 2016 17:44 EST
Performance impact because of Pega 0011 alerts
Our apps have been migrated from Pega 5.5 to 7.1.8.
One of the pega (7.1.8) app servers is configured to take SOAP requests from the front end, there are many PEGA0011 alerts (with response times going to 50+ seconds sometimes), this alert just gives the information about the Service that took more time to respond and no other details.
We couldn't see any other alerts (no database alerts, no external connection alerts..) that we could use to find the reason for the slow response time of the service hosted in pega.
Can someone provide suggestions on how to analyse further in this case?
Regards,
Hareen
Pegasystems Inc.
US
You can start with this: https://community.pega.com/sites/default/files/help_v719/procomhelpmain.htm
There is also SOAPService class debug log you can turn on to identify the timeline. If you issue is robust, this should give clues what is responsible.
Pegasystems Inc.
IN
Can you check on the overall network connectivity?
you may want to review below article as well:
https://community.pega.com/sites/default/files/help_v719/procomhelpmain.htm
Pegasystems Inc.
GB
Hi Hareen,
Do you know the rate of incoming SERVICE calls being made ? I mean : how many requests per minute ?
What are the settings for your Requestor Pool ? I'm not sure: whether the PEGA0011 include or exclude 'waiting time' here or not: but assuming for the moment that they do: perhaps your incoming requests are spending time waiting for a free Requestor to be made available in the Pool ?
You can use SMA to monitor the Requestor Pools: to see if this is a possiblity - you can also check your logs for any evidence of any explicit timeouts ('Timed Out Waiting for Requestor' I think the message is).
I would NOT just increase your Requestor Pool setting without first monitoring them - because you have to consider the trade-off between time waiting and resources (memory mostly) used.
Cheers
John
Pegasystems Inc.
IT
Did you changed the default threshold for alerts?
If so you can restore the original values or even decrease them just for the test period.
Then you may want use your preferred testing tool (SoapUI or whatever) to load test your service in order to compare performances observed from the outside with the ones collected from the inside.