Possible SNMP issue

[davidbaker]

We can not see your pictures 2 posts up.

UPLOAD the pictures to the forum with the link below the BLUE submit button.

Fixed. I edited the post and uploaded the files instead of providing links.

[davidbaker]

Any new updates from the Netonix side?

[Eric Stern]

As a first step I'd suggest you need to figure out why the switch can't reach the NTP server. The status page shows that the date is not being set correctly.

[davidbaker]

As a first step I'd suggest you need to figure out why the switch can't reach the NTP server. The status page shows that the date is not being set correctly.

Right....... but the NTP has been an issue. That wouldn't explain why after we updated the firmware my interface speeds keep reporting something different with every poll. Also the last boot date never changed until the update. The switch appears to be sending false SNMP measurements since updating firmware.

[davidbaker]

Okay, I fixed the NTP issue (put in a local NTP on the subnet) but I still have the same problem. SNMP measurements are still not reporting correctly. Also CPU is still spiking to 99%... I really dont want to have to downgrade but since I fixed the original issue that caused us to upgrade the firmware I may have to unless you guys have any other suggestions.

[sirhc]

I would like you to try v1.4.5rc8 which is the latest version which may help the CPU issue.

If "after upgrading to v1.4.5rc8" the CPU does not calm down after about 10 minutes try disabling services under the Device/Configuration Tab one at a time until the CPU drops.

Things like:
SMTP
Discovery Protocols - UDP, CDP, LLDP
Discovery Tab
TFTP Backup
SYSLOG

The idea would be to determine what is causing the CPU load keeping in mind that the CPU load has nothing to do with the switching of packets as the CPU runs the UI and DAEMONS like SNMP and such and the core does the packet switching and can handle line speed.

Another thing would be to look at how often your SNMP server is polling the switch for information, make sure your not polling it too often. I have seen people set their servers up to poll the switch every second which is silly.

Also if your using our Manager and it is polling the switch make sure the Manager is also upgraded to the last version and even try telling the Manager to stop monitoring the switch.

The idea is to debug the situation and determine what is causing the issue.

[davidbaker]

I would like you to try v1.4.5rc8 which is the latest version which may help the CPU issue.

If "after upgrading to v1.4.5rc8" the CPU does not calm down after about 10 minutes try disabling services under the Device/Configuration Tab one at a time until the CPU drops.

Things like:
SMTP
Discovery Protocols - UDP, CDP, LLDP
Discovery Tab
TFTP Backup
SYSLOG

The idea would be to determine what is causing the CPU load keeping in mind that the CPU load has nothing to do with the switching of packets as the CPU runs the UI and DAEMONS like SNMP and such and the core does the packet switching and can handle line speed.

Another thing would be to look at how often your SNMP server is polling the switch for information, make sure your not polling it too often. I have seen people set their servers up to poll the switch every second which is silly.

Also if your using our Manager and it is polling the switch make sure the Manager is also upgraded to the last version and even try telling the Manager to stop monitoring the switch.

The idea is to debug the situation and determine what is causing the issue.

Okay no problem I can try that.

I can say that our server is polling every 2 minutes for detailed stats (Interface info) and every 5 mins for node stats (CPU, Temp, ect..)

Ill let to know once what happens after the upgrade.

[sirhc]

Yea, also play with the polling rate to see if that maybe the issue.

[Nitroxide]

I have this same issue with solarwinds NPM and these switches.

[davidbaker]

Yea, also play with the polling rate to see if that maybe the issue.

I changed the polling rate to every hour, interface speeds still jumping all over the place.