Wired Intelligent Edge

We have a switch in a closet deamed 'too small' to install an AC. The room stays 79-84 degrees for the most part since summer has began. 

 

I know the product documentation says up to 122 degrees. Is that accruate? How much performance degradation should be seen at higher temperatures?

 

All of our other switches are in rooms around 70 degrees (they all have AC). We are hearing complains from the floor that this switch is on, and we are trying to decide if it's because of the temperature. 

 

 

This isn't a direct answer to your question but I've had 8 switch stacks in closets with no AC and room temperature hovering in the high 90s. No issues.

I personally can't see a 'huge' problem with high temp like that. I mean I've had other vendor  switches in rooms hotter, but we're trying to figure out where this 'issue' is coming from that our users are reporting. 

If the unit was in danger of thermal overload, it would actually shut itself down. Typically exceeding thermal thresholds affects the lifespan of the product, not degrade performance during the event.

 

What does show inventory show with respect to fan speeds? The output would look like this:

 

System Fan Tachometers  : Power Supply Fan 0      : 9507  RPM (Normal Speed)
                        : Power Supply Fan 1      : 9963  RPM (Normal Speed)
                        : FAN 0                   : 5142  RPM (Low Speed)
                        : FAN 1                   : 5162  RPM (Low Speed)

 

Unless the fans are running at HIGH SPEED, i wouldn't necesarrily be concerned with temperature.

 

Could you describe the issues the users are reporting?

 

I rarely see the fans running High. It doesn't even happen daily. Perhaps once a week the fans will run high for about a half hour. 

 

Power Supply                  : Present (580W)
                              : 12V System Voltage Ok
                              : 56V PoE Voltage Ok
System Temperature            : CPU Temp                : 54  C
                              : DPE-0 Temp              : 71  C
                              : DPE-1 Temp              : 67  C
                              : DDR Temp                : 46  C
                              : PSU Temp0               : 40  C
                              : PSU Temp1               : 43  C
                              : Exhaust Temp            : 47  C
                              : Mid Temp                : 54  C
                              : Ambient Temp            : 45  C
                              : Console Card Temp       : 30  C
System Voltages               : 3300mV Sense            : 3376.0 mV
                              : 2500mV Sense            : 2464.0 mV
                              : 1800mV Sense            : 1824.0 mV
                              : 1000mV Sense            : 1008.0 mV
                              : 1100mV MAC              : 1092.0 mV
                              : 1200mV Sense            : 1216.0 mV
                              : 5000mV 5VDD             : 5022.0 mV
                              : 5000mV 5VSB             : 4974.0 mV
                              : 900mV VTT               : 922.0 mV
System Fan Tachometers        : Power Supply Fan 0      : 4736  RPM (Low Speed)
                              : Power Supply Fan 1      : 5037  RPM (Low Speed)
                              : FAN 0                   : 5075  RPM (Low Speed)
                              : FAN 1                   : 5094  RPM (Low Speed)

Here is power and Fan for that switch.

 

The users (all users connected to this switch, and only this switch) complain of slow/laggy internet. VOIP call quality containing static at times, and buffering on low definition videos. 

 

I am mostly concerned with the phones, but obviously I'd like to ensure a good overall connection experience. If it was 1 user I wouldn't suspect the hardward, but with all of the users on that floor complaining, there must be a common factor. 

If the system was struggling with the temperature of the closet, I would assume you'd see the fans run HIGH more constantly than once a week or so. What version of AOS is this switch running? FANs can run high not only out of necessity to keep the system cool, but also based upon PoE load. When this HIGH event accurs, is there additional PoE consumption on the switch (e.g. greater wireless/phone traffic which causes the endpoint to consume more power)?

 

Also, have you checked interface counters for example to see if you're taking errors on links? What about any potential STP activity (e.g. ports thrashing)? 

 

How quickly are you notified by the users that they are experiencing issues? Can you capture the logs during the event?

 

Best regards,

 

Madani

I agree. I really don't think it's the temperature in the closet, but that's the only factor that I can see that makes this switch any different from our other closets. 

 

The switch is running AOS 7.3.1. 

This switch is also only half populated. It is a 48 port switch, we are using 26 ports at this time.

All of the PoE phones are stationary and they are rarely unplugged, so the amount of PoE draw should be relatively steady. 

It is possible that there is greater traffic at times, this is an administrative floor where not everyone is Always in the office, so that could play a role in why this is hard to diagnose. 

 

The counters available on the GUI show Zero events. Airwave also does not indicate an issue. 

 

 

This switch is a connection point between our two stacks though. We have a fiber link between a stack with 3 switches and this stack. All traffic comes in on a single vlan as it is all AP-93Hs. The stack connected to this is our 'dorms'. 

 

 

 

Can you private message me your logs package? I'll take alook around...

This may be a novice request but. 

 

I cannot for the life of me find a location in the user guide or anything that explains how to export the logs. I can export the Config, but I don't see a logs file for the mobility switch stack.

In the WebUI of the switch, you can go to the maintenance tab and download the logs

 

So a large source of my issue seems to have stemmed from STP related issues. 

 

I have made some configuration updates with the help of Madjali and now we play the sit and wait game to see if my users still run in to any issues.

 

 

Thanks for your help today Madjali!