WS-26-500-DC issues when SNMP enabled

[CrackerRiley]

When I enable SNMP, CPU usage is above 90%, reported total throughput thru GUI is 4x more than actual throughput, "i2c thread" process higher than normal load average.
Same behavior when using 1.4.8rc and 1.4.9 now.
SNMP has been turned off and behavior still happening.

We are monitoring:
Fans speed
Temps
Input volt
total wattage
CPU usage
uptime

Load avg continues to rise the longer SNMP is enabled. Currently at 6 after a few minutes...
What do?

[sirhc]

GUI throughput is accurate. However overall throughput counts through traffic twice as it comes in and then out on 2 port so 2X unless you uncheck the box on the Ports TAB to not count that port towards total throughput


CPU usage high probably because your polling switch too often.
viewtopic.php?f=6&t=2723&p=18877&hilit=+snmp+cpu#p18877
viewtopic.php?f=6&t=3350&p=22105&hilit=+snmp+cpu#p22105

[CrackerRiley]

Hmm I get the 2x throughput deal, that's not the case here. I can click on the port that uplinks to our router and get this:

This is minutes after reset all the counters. I know 10GBs hasn't gone thru that port yet..

At the same time our router shows this:

SNMP is no longer even enabled, but when it is we are polling every 4 minutes.

I'm the only one logged into the switch and CPU usage has never dropped below 90%. Reporting seems to return to normal when the switch has been rebooted but this is one of our busiest sites..

We have 2 other 26-DCs on our network with the same setup that don't seem to be having these issues (although their CPU usage is always around 90% as well).

[sirhc]

The CPU utilization on the WS-26-500-DC is normally higher than any other model as the little MIPs CPU that runs the UI/CLI , stats daemons and other functions like reporting to SNMP, discovery, and so on.

You can disable feature on the Device/Configuration TAB and see what feature is using the CPU.
Disable things like SNMP, SMTP, Discovery and so on and see how each affects the CPU

Keep in mind this CPU has nothing to to with switch functions like packet forwarding and has nothing to do with load or capacity.
https://forum.netonix.com/viewtopic.php ... pu+#p22102

As far as the port reporting bandwidth passed has been investigated and verified many times? Since all the switches use the same switch core, same firmware, and so on the issue would be the same across all models if it existed? The switch core keeps track of these numbers and we simply pole and report these values and the switch core is a mature product and there are no bugs. That part of the code is written by the switch core maker not us.

If you look at the 1 hour graph for bandwidth it is a better representation of average throughput as you may be missing MANY spike that may have occurred that skew the bandwidth unless you sit there and watch that 30 second graph and pay close attention. But then again the 1 hour graph is a better representation of the overall average.

You can go to the CLi and drop to Linux shell and issue a "TOP" command and see where the CPU usage is being utilized and post a screen grab of the TOP command.

So to do this
Goto to CLI which you can do in the web UI or ssh
type "cmd" and press enter to drop to Linux shell
type "top" and press enter
do a screen cap and post it here.