Loading...
Loading...
- Periodic CPU spikes observed on odd-numbered CPU cores within a 36-core tenant environment - MCPD is consuming approximately 100% CPU over sustained periods
- MCPD main thread consumes ~100% of one CPU core continuously for the duration of the affected uptime - CPU appears as periodic spikes on individual cores (system mode dominant) - No MCPD restart or traffic outage occurs spontaneously. The system remains operational but with degraded CPU resources.
BIG-IP TMOS v21.0.0 (multi-threaded mcpd) is deployed on VELOS C128 and rSeries r10k platforms. In an HA pair or device group with configuration synchronization, the peer socket receives an EOF while MCPD has outstanding requests greater than zero on that connection, which may occur during or after a configuration sync. This issue has been observed in a 36-core tenant configuration and during extended system uptime
Restart mcpd to clear the orphaned file descriptor and restore normal CPU utilization bigstart restart mcpd Note: Restarting the MCPD causes a complete traffic outage on the affected unit In HA deployments, perform a failover to the standby unit before restarting mcpd on the affected active device to minimize traffic disruption
None
F5 Integration
Learn more about where this data comes from
BugZero Plan
Streamline upgrades with automated vendor bug scrubs
BugZero Prevent
Wish you caught this bug sooner? Get proactive today.