High or 100% CPU seen in the system status graph may indicate a problem. Administrators should be concerned if:
- CPU is constantly increasing over time and never decreasing.
- CPU is consistently high with little or no load on the device.
- Slow performance of PCS including users experiencing slow login time, slow response through the VPN tunnel and admin ui slowness.
- Cluster nodes consistently loss connectivity which cause cluster splits or VIP fail overs.
Normal CPU behavior are the following:
- CPU should increase as user count and throughput numbers increase.
- CPU should decrease as user count and throughput numbers decrease.
Under these conditions, if CPU does reach 100%, these are signs the device may be reaching their hardware limit. For further confirmation, please gather the logs and open a support case at https://my.pulsesecure.net
Logs to gather for root cause analysis:
- System snapshot during or near the time of the issue. If any process snapshot, these should gathered as well.
- PCS logs
- Screenshots of all system status graphs (covering the time of the issue)
- Please note the time of the issue
- Navigate to TroubleShooting > System Snapshots
- Click Take Snapshot
- After a few minutes, a link for Admin generated snapshot will appear. Click the link to download the snapshot. If any additional process snapshot appear, download these files as well.
- Navigate to System > Status > Overview
- Ensure the graphs displayed covered the times of the issue, then take a screenshot(s) of all graphs
- Navigate to Logs/Monitoring > Events > Save All Logs
- Open a support case at https://my.pulsesecure.net
WHAT CAN AN ADMIN DO?
- Is debuglog enabled with excessive detail level for extended period (either left enabled accidentally or done per support team recommendation)? If debuglog is enabled, turn debuglog off.
- Is someone running a log query, specially on a large user access log, during heavy PCS usage? If log query is configured, remove the log query and confirm if CPU does decrease.
- Is archiving running? Archiving can cause short CPU spikes. Once the archive is complete, cpu should decrease.
- Are there any event logs for process crashes or critical messages? Please note process crash or critical messages when opening a support case.
- Are there any process or watchdog snapshots being created? Download all process or watchdog snapshots from the System Snapshot page and include these as attachments when open a support case.
- Call Pulse Secure Support and fully describe user experience and the changes (if any) as per above examples. Submit all logs.