A common issue encountered with ControlUp dashboards involves inaccurate reporting of process start times and user counts, specifically with high numbers of users and processes that initiate immediately after login. In a troubleshooting case, a community member observed that the Netskope client was not routing traffic to on-premises file shares for 5 to 10 minutes post-login. They wished to confirm through a dashboard or report that the service in question (stAgent process) started almost immediately after user login. However, the dashboard filtered to this process was showing many "N/A" values for relevant metrics, and the total user count was about 900 fewer than expected.
The core technical challenge was identified as a data collection nuance related to where process data originates and the timing of process starts relative to the SIP (Session Initiation Protocol) agent. It was clarified that the metrics came from "process stops" indices. Users noted that usernames were sometimes missing from these entries because the process starts before the SIP agent logged the user identity, leading to incomplete user data capture. Additionally, the "app_launch_time_ms" metric was missing for some process start events, likely due to this timing discrepancy. Differences were highlighted between two different data sources, "process_stops" (Windows-focused) and "builtin_process_stop" (Mac-focused), explaining variations in available metric details.
Another topic discussed was the limitation encountered when interacting with Stat widgets in ControlUp dashboards. When clicking on a stat card representing many devices (e.g., 2000 unique device IDs), the drill-down view only displayed up to 500 devices, causing confusion and incomplete visibility. This limitation was acknowledged as potentially unintentional and slated for further investigation. The community debated the practical need for disabling this drill-down feature in some dashboards, with consensus that although direct action on large device sets might be unusual, users still expect full visibility matching the dashboard summary counts.
From a solution perspective, filtering widgets by exact username with a negation filter was suggested to show only entries with recorded usernames, assisting in troubleshooting user data completeness. The issue with incomplete metric values related to early-starting processes remains a technical limitation tied to data collection timing, without an immediate fix but with suggested workarounds for filtering and interpreting available data.
In summary, to troubleshoot processes that start immediately after login but show incomplete metrics or user data in ControlUp dashboards:
- Understand the difference between the Windows "process_stops" and Mac "builtin_process_stop" indices and their data characteristics.
- Recognize that processes starting before the SIP agent registers the user can result in missing usernames and launch time metrics.
- Use filtering techniques to isolate entries with valid usernames.
- Be aware of drill-down limits on Stat widgets (currently capped at displaying 500 devices), which affects visibility on large datasets.
- Monitor ControlUp updates for potential removal or adjustment of drill-down limits.
For more details on dashboards, filtering, and process monitoring in ControlUp, refer to the official documentation at https://docs.controlup.com, and for dashboard customization best practices, visit https://cuacademy.controlup.com.
Read the entire article here...
ControlUp Community
Connect, Learn, and Grow