Loading...
Loading...
An issue has been observed in the Yokohama and Zurich releases where the Analytics Center NLQ service may generate excessive scans when users interact with the “Ask a question about your data” in the “Analytics Center.” The logs of such user interactions are available in the `nlq_query_log` table, where: The origin parameter is set to AC, and The table parameter is empty The processor handling such queries may cause Java heap space Out-of-Memory errors, resulting in application node restarts, temporary hangs, and instability while producing responses to analytical queries in the analytics center due to high memory usage.
Log in to a ServiceNow instance. Navigate to Platform Analytics → Analytics Center. In the “Ask a question about your data” input field, enter a simple query such as "closed". Click Ask to execute the query. Observe that the NLQ service issues an unbounded query because the origin parameter is set to AC and the table parameter is empty, which triggers scans across extremely large datasets (e.g., ~10 million CMDB records). The system continues processing excessive scans, leading to prolonged execution times and ultimately may lead to Java Out‑of‑Memory errors, which may cause node instability or restarts.
The problem has been addressed in the versions listed in the Fixed In section below. This issue has been addressed in Yokohama Patch 11, Zurich Patch 5, and later releases, where customer may choose to disable the NLQ support for Performance Analytics (PA); however, manual action is required after the upgrade to apply this workaround.
Options: Option 1: For instances on Yokohama Patch 11, Zurich Patch 5 or later releases and affected by this issue. Apply the workaround given below: Disable NLQ support for Performance Analytics (PA) by setting the Boolean system property: com.snc.par.nlq.pa.enabled to false. After updating the system property, monitor NLQ activity and overall system behaviour to confirm that resource usage is reduced and memory-related issues are resolved. Option 2: For instances on releases prior to Yokohama Patch 11, Zurich Patch 5 and affected by this issue. Apply the workaround given below to reduce the risk of your instance experiencing this issue: Create an integer system property with name com.snc.par.nlq.max_breakdown_elements_from_none_sys_choice_tables, and Set it’s value to 10 (Recommended value: 10 or lower. The default value without this property is 256). It limits the number of breakdown elements processed and helps stabilize requests. After updating the property: Run: cache.do In the sys_analytics_schema_cache table: Set evict = true for all records (Typically, only a few records per language) Re-run the NLQ request to regenerate cache entries Notes: This system property only affects the metadata about the Performance Analytics Breakdown Elements sourced from non-sys_choice tables in the Analytics Center Q&A feature. It does not affect PA data collection, other PA features or UIs, or NLQ functionality elsewhere on the ServiceNow platform If request timeouts still occur in BreakdownMappingsMetadataGenerator after setting a low value, you can: Set the system property to 0, or Manually create empty cache entries: For the following records in sys_analytics_schema_cache table: breakdown_metadata_en breakdown_mappings_en Set: schema = [], evict = false, Increment the version number by 1. After making changes, run cache.do to ensure the updates take effect Additionally, please note that this feature is deprecated in the Australia release. FAQs: Q. Can the customer implement the workaround themselves? A. Yes, the workarounds mentioned above can be implemented by the customer by following the instructions provided Q. Is there a permanent fix available for this issue? A. No, there is no permanent fix for this issue. Customers experiencing this issue may choose to apply the workaround(s) mentioned above to prevent this issue from affecting their instance.
PRB1917953
Click on a version to see all relevant bugs
ServiceNow Integration
Learn more about where this data comes from
Bug Scrub Advisor
Streamline upgrades with automated vendor bug scrubs
BugZero Enterprise
Wish you caught this bug sooner? Get proactive today.