problem with SG200-08

[jknockaert]

Once in a while my SG200-08 switch will not recognise new devices connected to it. It simply will fail to update the status of the ports, so when in such a state the green light will stay on after disconnecting a device, and it will not establish a new connection after (re)connecting a device. It will however continue to switch traffic for devices that have been successfully connected before the issue started. The only solution to the issue is to restart the switch.

I could identify steps to reproduce the issue, but it happened with all firmware versions so far (I'm now using 1.0.5.1). I'm not sure what triggers the behaviour. Once I could observe that the switch was very slow in establishing a new connection, and after that it would not update the port status anymore (so I had to restart it).

[Tom Watts]

Hi Jasper,

A few things I can think of that may help-

• Check the spanning-tree interface properties and see if port fast (fast link) is enabled
• Remove EEE (energy efficient ethernet) globally and per port
• Try to set the port in question to manual negotiate 100 full or 1000 full instead of auto

-Tom
Please rate helpful posts

[jknockaert]

Hi Tom

Today the switch went into numb mode again. It was generally slow, both in negotiating new connections and in the user interface. Negotiated settings for new connections were however fine on both sides. So I digged a bit further, and found in the diagnostics section that the cpu utilization was 99% (5 minute average). I couldn't find any indication on what is causing the issue, but after a reset of the switch the cpu is again running at a more appropriate 5%. Is there a way to find out what process is hugging the cpu?

Jasper

[Tom Watts]

The switch switch has a flash log. The unit won't tell you what is spiking the utilization but the flash log will indicate any serious errors.

-Tom
Please rate helpful posts

[jknockaert]

The flash log is empty. I also checked the RAM log before rebooting the switch, and there was nothing exceptional there. Log settings are the default ones, i.e. everything but debug messages are logged in RAM. Any other suggestion for identifying the cause of the high cpu utilization?

[Tom Watts]

Try to check each individual interface under status and statistics. See what port is sending/receiving the most traffic then I'd start with those units.

-Tom
Please rate helpful posts

[jknockaert]

Well, I'll have to wait for the switch to run into trouble again to check it's port statistics upon that event. But I did a check on the devices connected to it, which is basically a server and an uplink. The server reported something like 4KB/s network troughput with few spikes in the hours before I diagnosed the issue with the switch. The uplink modem reported under 100kbit/s DSL throughput in each direction (which is consistent with the server measurement given the resolution of the modem measurement and accounting for the DSL overhead). Also did the lights on the switch not reveal any particular network activity. I will report back when I have more observations.

[jknockaert]

Ok, today the switch ran again into cpu hogging mode. The diagnostic indicated 99% cpu usage over the last 5 minutes. This time I disconnected all devices, but let the switch powered and didn't reset it. More than two hours later I reconnected a laptop and immediately checked the diagnostic. It told me that cpu usage (5 minutes average) was still at 99%. There seems to be a real issue here that is not caused by traffic intensities but rather by a runaway process. Any other suggestions for diagnosing?

[Tom Watts]

Hi Jasper, try to give it a firmware flash and reset the unit (I am aware you are using the 1.0.5.1). If that still doesn't settle it down, give the SBSC a call. Something isn't quite right there.

http://www.cisco.com/cisco/software/release.html?mdfid=283454003&catid=268438038&softwareid=282463182&release=1.0.5.1&relind=AVAILABLE&rellifecycle=&reltype=latest

-Tom
Please rate helpful posts

[jknockaert]

Ok, I flashed the firmware and reapplied a minimal set of customisations (such as fixed ip and ntp server), the rest of settings is factory default. Today the switch again run into cpu hogging mode. I'll check how I can get into contact with SBSC. Jasper

[stianlie1]

Hi Jasper,

I had the same issue with two SG200-08, one PoE and one standard, both running 1.0.5.1 firmware. They both showed 99% processor load constantly.

I tried a bunch of things, but for me "Administration/Discovery - Bonjour" and unchecking Enable (disabling it) sorted this out on both. Based on this I would think that there is an issue with the implementation of this feature.

Regards,

Stian

[jknockaert]

Hi Stian

Thank you for your feedback. After my switch (a standard one) got replaced (by a PoE one), the new switch revealed the same behaviour of running at 99% cpu load after about one month. I then switched Bonjour off (that was early January), and since then I didn't experience the 99% cpu load event. So it may be gone; however: if the reported uptime is correct, the switch has been resetting itself spontaneously a couple of weeks ago. Not sure why, there hasn't been a power failure on that circuit. Anyway, a new firmware has been released this week.

Jasper