Identifying high transaction concurrency - Support and Troubleshooting

Identifying high transaction concurrency Description  Identifying high transaction concurrency

Symptoms:

Slow response times. Instance appears to be inaccessible. Stats and threads pages are accessible, but the user interface is not Cause Too many users based on instance sizing. Integration traffic overwhelming transaction capacity. Resource constraints slowing down transaction time. Resolution To verify if concurrency issues exist:

Gather and save stats.do and threads.do for the instance.

Note : If you are unable to reach the stats.do page, a restart of the instance or individual node may be necessary. Contact the ServiceNow Technical Support team to determine if this is necessary to perform the action. Refreshing the stats.do page will give you an idea of concurrency over time. Observe the Semaphore Sets section on the stats.do page over a few refreshes. The instance may be experiencing a high concurrency of requests if: Available semaphores is at 0 (zero) , or remains very close to that. Currently waiting on a semaphore remains at 1 (one) or above . Semaphore transaction times are low (sub 1 second) , but the above conditions are still true. If semaphore transaction times are long, then this is not a concurrency issue and is likely a blocking or other performance issue. If the long-running semaphores are mostly for the same URL (for example, /incident.do ), then there may be an issue related to that individual page, and further troubleshooting of that page is needed. If the long-running semaphores are for various kinds of URLs, then there may be a resource constraint causing degraded performance. For additional information, see Troubleshooting slow performance . Alternate verification method:

If you are able to log in to and use the instance, navigate to the Service-now Performance homepage. Observe the System Overview , Response Time , and Transactions graphs for the relevant time period. The instance may be experiencing a high concurrency of requests if the graphs display any of the following scenarios: System Overview : The bar graph representing concurrency_mean is greater than or near the number of maximum semaphores. Response Time : The Median line on the graph is in the acceptable range (sub 1 second) . Transactions : Transactions per minute are high. For most systems, 200K per minute and above are worth investigating. If transaction times are long or transactions per minute are relatively low, then this is not a concurrency issue. Other possible causes of performance issues should be investigated.

To resolve a concurrency issue:

Suspend activity at the source of the traffic, if possible. If this is a Management, Instrumentation, and Discovery (MID) Server or external integration, the owner of the system will need to take action.

Note : To help identify the source of the increased request traffic, contact the ServiceNow Technical Support team. If transaction volume is related to normal user traffic, an incident must be created and assigned to Technical Support to evaluate instance capacity.

If the issue continues to exist after following these guidelines:

Clearly identify the issue or question. Visit the ServiceNow product documentation . Search the ServiceNow Community . Post a question on the ServiceNow Community forums . New users must create an account on the ServiceNow Community in order to post. Open an incident via email or the online Technical Support system. Contact the Technical Support team. Additional Information For more topic related issues, review the following Knowledge Base article:

Troubleshooting slow performance