Status Page Weirdness

[adairw]

See the attached picture. I am running 1.0.4 and while watching the status page I keep seeing all the voltages go to 0 or some negative number. the next update they will be back to normal and a few seconds later back to zero. I don't remember this with the very first version I loaded on the switch. (1.0.2RC-something)

[sirhc]

hmmm I am not seeing that but I have all 24 port model A's in service?

I will have to load up and test the 12 port in a little while.

Other than this is the new firmware working OK?

[Dave]

Can you go to the status page & see if the DC voltages are also jumping around?

[lligetfa]

I keep seeing all the voltages go to 0 or some negative number...

I think you mean wattages. I have not seen any go negative but I see fraction positive wattages showing on 48V hotted up ports with nothing plugged into them.

[adairw]

Other than this is the new firmware working OK?

Maxxwave SFP still doesn't work. (as reported on my other thread)

[adairw]

I keep seeing all the voltages go to 0 or some negative number...

I think you mean wattages. I have not seen any go negative but I see fraction positive wattages showing on 48V hotted up ports with nothing plugged into them.

Yes, wattage, of course. Brain was low on coffee.

[adairw]

Can you go to the status page & see if the DC voltages are also jumping around?

Yes, they are also bouncing around.

WS Voltage.PNG (3.04 KiB) Viewed 7656 times

Just for info, this is Chrome Version 38.0.2125.122 on Windows 7 Pro

[Dave]

I think your SFP module has a bad I2C bus. We have noticed that often when a SFP module goes bad, its network ability can still work, but its I2C function acts weird. The SFP modules are only suppose to respond to a certain I2C address, yet a bad SFP will respond to address at random, which is why you see the voltages/power jumping around. It is interfering with our I2C A/D readings.

I purchased a bunch of SPF's on line at a big discount & when they arrived (I was told they were used but still work) I found that they all had bad I2C buses. Used a new known good one & everything worked fine.

Any chance you can try a different SFP module?

[adairw]

I think your SFP module has a bad I2C bus. We have noticed that often when a SFP module goes bad, its network ability can still work, but its I2C function acts weird. The SFP modules are only suppose to respond to a certain I2C address, yet a bad SFP will respond to address at random, which is why you see the voltages/power jumping around. It is interfering with our I2C A/D readings.

I purchased a bunch of SPF's on line at a big discount & when they arrived (I was told they were used but still work) I found that they all had bad I2C buses. Used a new known good one & everything worked fine.

Any chance you can try a different SFP module?

I think the only other one I have is a maxxwave SM fiber SFP but I only have one.
I will pull the sfp and see if the switch starts acting normal again. I'll then plug in the SM SFP and see what it does also.

Keep in mind, these SFP's worked fine when connected to both mikrotik and hp procurve devices. I used these SFP's for a long time at home just testing them out. not saying there isn't a problem, but I do find this odd.

[sirhc]

The problem comes when an SFP module answers request on more addresses than it is supposed to.

SFP modules have an address of 50 on the I2C bus and we are using the address 51 for the current and voltage sensors.

The SFP module will work fine as far as the media or data is it is passing is concerned so if you would use this SFP module in a switch that is not sharing the I2C bus to do different things like we are for current sensing you would never notice it miss behaving.

So whats going on here is that when we are requesting info from the current sensors on the I2C bus at address 51 the SFP module is answering when it should not be which is why you see wrong answers on the current and voltage sensors.

And Les those phantom point whatever small current levels you see on ports with their POE turned on but nothing there is because when we calibrated the current sensors Dave had to chose what range it be more accurate for our application and since we are not looking for really small .01 watts it sometimes see phantom loads. I guess Eric could put in a kludge to ignore those low loads.

If I am explaining this incorrectly on the I2C or phantom loads then I am sure Dave will correct me here.