[evla-sw-discuss] MCCC Kernel upgrade and modification

[James Robnett]

Done.  The current setting is the max 511.  The default was 200.

One of the cool things about setting that value is it causes the 
interface to replumb which effectively resets the stats.  So the framing 
or dropped packet counters are reset to zero.


[root at mccc ~]# ifconfig eth0
eth0      Link encap:Ethernet  HWaddr D4:AE:52:B6:5D:F6
           inet addr:10.80.100.109  Bcast:10.80.100.255  Mask:255.255.255.0
           UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
           RX packets:67988 errors:0 dropped:0 overruns:0 frame:0
           TX packets:81907 errors:0 dropped:0 overruns:0 carrier:0
           collisions:0 txqueuelen:1000
           RX bytes:16449206 (15.6 MiB)  TX bytes:8863024 (8.4 MiB)
           Interrupt:177 Memory:d90a0000-d90b0000

[root at mccc ~]# ethtool -g eth0
Ring parameters for eth0:
Pre-set maximums:
RX:             511
RX Mini:        0
RX Jumbo:       0
TX:             511
Current hardware settings:
RX:             511
RX Mini:        0
RX Jumbo:       0
TX:             511


On 04/08/2014 08:07 AM, James Robnett wrote:
>
> Fair enough.  I'll make the change as soon as I hear we're not
> observing.  It shouldn't be disruptive but it could pause traffic
> momentarily.
>
> Is there any chance of a script being run today that might trigger
> the dropped packets (either fast configurations or delay models).
>
> I'd characterize the RX buffer issue as speculation.  It fits the
> symptoms but I wouldn't assume it's definitely going to fix things.
>
> james
>
>
> On 04/07/2014 04:37 PM, Michael Rupen wrote:
>> With tomorrow the first of two maintenance days, and with one arm out,
>> I'd say you should think about just playing games with mccc itself on
>> Tuesday.  Unless that's painful for some non-operational reason.
>>
>>            Michael
>>