ping: sendto: No buffer space available

[Garrett Cooper]

On Mon, Sep 24, 2012 at 8:06 PM, Rudy (bulk) <crapsh at monkeybrains.net> wrote:
> On 9/24/12 5:01 PM, Ryan Stone wrote:
>>
>> Can you get the output of netstat -I emX -d?
>
> ...
>
>>
>> I suspect that you are seeing the em TX queue fill up.  If so you
>> should see output drops reported by the em interface.
>>
> I do see 205 in the 'Drop' column.  I ran the command twice two hours
> apart... I am seeing Ierrsincrement in a 2 hours window...
>
> #date && netstat -I em1 -d
> Mon Sep 24 17:47:14 PDT 2012
> Name    Mtu Network       Address              Ipkts Ierrs Idrop    Opkts
> Oerrs  Coll Drop
> em1    1500 <Link#2>      00:25:90:26:62:01 110170687801 4738950     0
> 85411108876     0     0  205
> em1    1500 X.X.X.X       coconut.coconut-g 111858     -     -  6536889
> -     -    -
> em1    1500 fe80::225:90f fe80::225:90ff:fe        0     - -        0     -
> -    -
>
> # date && netstat -I em1 -d
> Mon Sep 24 19:48:44 PDT 2012
> Name    Mtu Network       Address              Ipkts Ierrs Idrop    Opkts
> Oerrs  Coll Drop
> em1    1500 <Link#2>      00:25:90:26:62:01 110723416486 4768451     0
> 85861866459     0     0  205
> em1    1500 X.X.X.X coconut.coconut-g 119244     -     -  6558411     -
> -    -
> em1    1500 fe80::225:90f fe80::225:90ff:fe        0     - -        0     -
> -    -
>
> Checking all the interfaces, there are a lot more drops/Ierrs on em2...  The
> igb devices (PCIe card) seem a lot better than the Supermicro motherboard em
> devices.  Is this an on-board vs mb thing, or em vs igb thing?
>
> # date && netstat -id
> Mon Sep 24 19:53:49 PDT 2012
> Name    Mtu Network       Address              Ipkts Ierrs Idrop    Opkts
> Oerrs  Coll Drop
> em0    1500 <Link#1>      00:25:90:26:62:00 6887931271 844     0 18966954063
> 0     0    0
> em0    1500 X.X.X.X       coconut            1804123     -     - 1962796
> -     -    -
> em0    1500 fe80::225:90f fe80::225:90ff:fe        0     - -        1     -
> -    -
> em1    1500 <Link#2>      00:25:90:26:62:01 110745957108 4770615     0
> 85882005650     0     0  205
> em1    1500 X.X.X.Xcoconut.coconut-g   119573     -     -  6559700     -
> -    -
> em1    1500 fe80::225:90f fe80::225:90ff:fe        0     - -        0     -
> -    -
> igb0   1500 <Link#3>      00:1b:21:af:54:4e 53194278790 1169     0
> 85861440371     0     0    0
> igb0   1500 fe80::21b:21f fe80::21b:21ff:fe        0     - -        1     -
> -    -
> igb1   1500 <Link#4>      00:1b:21:af:54:4f 43905355772 0     0 24817874325
> 0     0    0
> igb1   1500 fe80::21b:21f fe80::21b:21ff:fe        0     - -        1     -
> -    -
> usbus     0 <Link#5>                               0 0     0        0     0
> 0    0
> usbus     0 <Link#6>                               0 0     0        0     0
> 0    0
> usbus     0 <Link#7>                               0 0     0        0     0
> 0    0
> usbus     0 <Link#8>                               0 0     0        0     0
> 0    0
> em2    1500 <Link#9>      00:25:90:26:62:02 2458414787 580252415422     0
> 1640052168     0     0 37284
> em2    1500 X.X.X.Xcoconut.guava-coc    93119     -     -   209407     -
> -    -
> em2    1500 fe80::225:90f fe80::225:90ff:fe        0     - -        2     -
> -    -
> em3    1500 <Link#10>     00:25:90:26:62:03 57160105 0     0 87941536     0
> 0    0
> em3    1500 X.X.X.XAS32329.weed-mb.c  1014267     -     -  1208726     -
> -    -
> em3    1500 fe80::225:90f fe80::225:90ff:fe        0     - -        1     -
> -    -

    I saw similar issues (although maybe not tied to e1000 directly)
when messing around with IPv6 interfaces (I have em and cxgb at my
immediate disposal right now).
Thanks,
-Garrett