Saturday, May 5, 2018

Been trying to figure out the cause of this for days..

Hi everyone!

I've been observing spikes in CPU usage on the MPC line cards we have within one of our companies Juniper MX960 edge routers and I'm really struggling to figure out what could be the cause of these spikes in usage as they appear to come out of almost no where (for the most part. I have observed a significant spike in usage to around 60% on the MPC's which occurred when over 1Mpps of TCP SYN packets was sent from a single sip).

However in this case I've checked the following from the minute the CPU usage spiked.

  • Interface in and out pps and bps statistics
  • Average packet size changes
  • Increases in unicast and multicast traffic
  • Increased activity in any interface filters

And so far I've come up empty trying to pin down the cause of this.

I've been using Observium for SNMP monitoring for quite a while now, and one of the only things that I can see that happens at the same time as the increase in CPU usage is that under ICMP statistics I can see a spike in out msgs and our errors as shown in the screenshot below. I can't see it being redirect messages as we have no-redirects and no-redirects-ipv6 set under the system hierarchy.

ICMP statistics - https://i.imgur.com/ILkdtlz.png MPC (MPC3-16x10GE) CPU usage - https://i.imgur.com/9joe008.png

Do any of you have any thoughts or suggestions to further diagnose the cause of this? Any assistance would be greatly appreciated.

Thank you.



No comments:

Post a Comment