[f-nsp] Log error on an RX-4 running 2703.a: bad i2c access

Dave Peters dave at terabitsystems.com
Tue Apr 26 12:44:49 EDT 2011


Thanks for all the responses.  Looks like this is a software bug.  It 
doesn't occur in 2.3, but does in 2.703a (I haven't checked the releases 
in between).  Severity minor.  I'm just going to let it go.

Thanks again, everyone.

On 4/25/2011 6:42 PM, Niels Bakker wrote:
> * dave at terabitsystems.com (Dave Peters) [Tue 26 Apr 2011, 00:40 CEST]:
>> Apr 25 23:14:47:I:System: bad i2c access (GIEI = set), Severity 
>> Minor, Mux 0x70, Channel 4, Addr 0x22 (FANTRAY1)
>> Apr 25 23:14:47:I:System: last good i2c access, Mux index 0x70, 
>> channel 5, Addr 0xd (PS2)
>>
>> This suggests the fan tray is bad, but a show chassis gives me the 
>> thumbs up.  Anyone have any experience with this?
>
> PS2 didn't release the I2C bus.  This is likely a bug in the firmware 
> of the power supply.  Brocade put some workaround code in later 
> versions for MLX, don't know about similar in RX but assume it's there 
> too, should be around the 2.7.02 timeframe, which you're beyond anyway.
>
> "Minor" I2C errors are recoverable and will generally only hang around 
> until the next "show chassis" or similar access is executed.  "Major" 
> I2C errors will require manual intervention to unstick the I2C bus. 
> It's annoying when that happens because the SNMP process hangs if 
> temperatures get polled, causing all snmp-based stats to go fubar.
>
> I've not seen "Major" error messages in several months now and only 
> the odd "Minor" ones, but those have become increasingly rare as older 
> hardware gets swapped out.
>
> If the issue happens frequently you may wish to ask Brocade for a new 
> power supply, assuming you have a support contract on the chassis.
> Remember to first unplug the power cord before removing the PSU.
>
>
>     -- Niels.
>


-- 
Dave Peters
Technical Director
Terabit Systems
2565 3rd Street  #218
San Francisco, CA  94107




More information about the foundry-nsp mailing list