Knowledge Search


[Junos Platform] VRR may report JTASK_SCHED_SLIP_KEVENT on multiple daemons

  [KB33953] Show Article Properties


With a Virtual Route Reflector (VRR) that is running on certain combinations of hypervisor software version and host machine hardware, you may see log messages indicating that multiple daemons, such as rpd/dot1xd/overlayd, are running into the JTASK_SCHED_SLIP_KEVENT error. This error may trigger a protocol timeout on MX Series routers.

This article gives the cause for the error to occur and a couple of workarounds.

Note: This problem is tracked in PR1411679.



When the VRR is syncing with the NTP server, you may see the following log messages and the protocol may timeout:

rpd[16192]: JTASK_SCHED_SLIP_KEVENT: 4 sec 898214 usec kevent block
dot1xd[16291]: JTASK_SCHED_SLIP_KEVENT: 5 sec 385737 usec kevent block
l2cpd[16245]: JTASK_SCHED_SLIP_KEVENT: 7 sec 582731 usec kevent block
overlayd[16296]: JTASK_SCHED_SLIP_KEVENT: 8 sec 313510 usec kevent block

Some known triggers for this issue are:

  • Periodic NTP sync due to NTP configuration

  • Executing the set date ntp CLI command



In this case, the VRR was selecting the HPET time counter that is provided by hypervisor by default. It is seen that some combinations of hypervisor software version and host machine hardware (CPU) provide inaccurate time count that confuses the NTP time sync handling under JunosOS/FreeBSD.



A permanent fix is expected to be made available in a future release. The issue is tracked in PR1411679.

Meanwhile some workarounds are as follows. Either one can mitigate the above problem.

  • Deactivate the NTP configuration and do not invoke set date ntp. This could be a temporary workaround for customers who do not want to reboot the VRR.


  • Add the following three lines to the /boot/vmguest.conf file in shell mode as the root user, and then reboot the VRR. 

Router>start shell user root
#vi /boot/vmguest.conf

Then add these three lines at the end of the file.



Related Links: