cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
442
Views
0
Helpful
1
Replies

SG300-28 Pegged CPU Latest Firmware

JamesRhyne29360
Level 1
Level 1

I apologize in advance as I am not the most familiar with networking and may be rough with some terms.

Recently, I acquired a SG300-28 for lab use and is in L3 mode primarily handling inter-VLAN routing across 9 VLANs, most traffic being on 3 of those, and a 4th VLAN that contains the default route to the internet facing router. Monitoring the device over the web GUI, SSH, and serial all show that CPU utilization is constantly >95% and regularly 99-100% even at night during lower traffic periods. It does seem to sit at 99-100% for 4 minutes then drop to 95-96% for a minute at a time but it is not always the same. At night, the router shows less than 1Mb of internet bound traffic from the switch and all the devices on it on average. I haven't noticed any considerable impact on the devices yet. The devices connected include 3 hypervisors that have interface bonds configured across 3 ports on the client side, no LAG or special configurations on the switch. All other devices are single port, nothing special.

Searching through here and other sites I have disabled Bonjour, disabled SNMP, toggled STP globally and on the ports individually, verified TCAM isn't maxed, and updated the boot code and firmware to the latest available but CPU usage is the same. Setting the logging threshold to debug has not shown any messages either.

 

CPU Input Rate: 10-15 f/sec avg

Firmware: 1.4.11.4

Boot: 1.3.5.06

 

Any help or pointers on what to check would be greatly appreciated.

1 Reply 1

JamesRhyne29360
Level 1
Level 1

To provide an update, a Prometheus deployment in a VM running Kubernetes is causing the CPU load to peak out on the switch. Stopping the deployment the CPU drops to an average of 10% as I would expect. We are still troubleshooting the issue for specifics on why it is causing it.