And - in OpenSolaris, it's bumped here:

http://src.opensolaris.org/source/xref/onnv/onnv-gate/usr/src/stand/lib/tcp/tcp.c#1006

:)

And - here -

http://docs.sun.com/app/docs/doc/817-6223/chp-mib?a=view

you can read about it in a little more detail from a dtrace perspective 
and what it means in a solaris context (pretty much the same(ish)).

And - keeping in the dtrace vein - if you have root access, which I 
suspect you will, you could do something like:

dtrace -n 'mib:ip::tcpAttemptFails { @agg[pid,execname,ustack()]=count(); }'

(All one line)

and see which processes are having issues, and what their stack looked like.

Then - if you were feeling really brave, you could perhaps consider 
playing with truss, dtrace or other friends, to work out what port # 
they are trying to do whatever they are trying to do it on.

Or - As you *should* already know *exactly* which ports should be in 
use... (heh heh) you can just look it up!

Nathan.

On  6/03/09 01:34 PM, A Gupta wrote:
> Looking at the Linux source tree, TcpAttemptFails is defined in
> include/net/snmp.h and incremented in net/ipv4/tcp_input.c,
> net/ipv4/tcp_output.c and net/ipv4/tcp_timer.c on a failed connection.
> I guess try tuning your TCP stack.
> 
> Cheers,
> 
> A
> 
> On Thu, Mar 5, 2009 at 8:26 PM, Chris Dimo <dimo at mail.org> wrote:
>> Folks,
>> can anyone explain in simple terms what "tcpattemptfails" ( netstat -s
>> ) means and what perhaps would cause this to increment around 10-15 units
>> per 30 seconds ?
>>
>> We have a situation where we moved servers into a new firewall complex and
>> suprise suprise, it is terribly slow and users have been complaining. ive
>> been running netstat, ping -s etc..and have the following results.
>>
>> qcmw002z# ping -s au1240s
>> PING au1240s: 56 data bytes
>> 64 bytes from au1240s (201.3.156.32): icmp_seq=1. time=2.91 ms
>> 64 bytes from au1240s (201.3.156.32): icmp_seq=2. time=3.62 ms
>> 64 bytes from au1240s (201.3.156.32): icmp_seq=3. time=3.74 ms
>> 64 bytes from au1240s (201.3.156.32): icmp_seq=4. time=3.08 ms
>> 64 bytes from au1240s (201.3.156.32): icmp_seq=5. time=3.25 ms
>> ^C
>> ----au1240s PING Statistics----
>> 6 packets transmitted, 5 packets received, 16% packet loss
>> round-trip (ms)  min/avg/max/stddev = 2.91/3.32/3.74/0.36
>>
>> -----
>>
>>
>> Thu Mar  5 20:14:08 EST 2009
>>         tcpAttemptFails     =144697     tcpEstabResets      =  4154
>>         tcpRetransSegs      =241399     tcpRetransBytes     =278717070
>>         tcpOutRsts          =223765     tcpOutFastRetrans   =   141
>>         tcpRttUpdate        =3633063    tcpTimRetrans       =136919
>>         tcpTimRetransDrop   =  3339     tcpTimKeepalive     = 33893
>>         tcpHalfOpenDrop     =     0     tcpOutSackRetrans   =186148
>>
>>
>> Thu Mar  5 20:14:38 EST 2009
>>         tcpAttemptFails     =144714     tcpEstabResets      =  4154
>>         tcpRetransSegs      =241400     tcpRetransBytes     =278717070
>>         tcpOutRsts          =223777     tcpOutFastRetrans   =   141
>>         tcpRttUpdate        =3633103    tcpTimRetrans       =136919
>>         tcpTimRetransDrop   =  3339     tcpTimKeepalive     = 33896
>>         tcpHalfOpenDrop     =     0     tcpOutSackRetrans   =186148
>>
>>
>> Thu Mar  5 20:15:08 EST 2009
>>         tcpAttemptFails     =144724     tcpEstabResets      =  4154
>>         tcpRetransSegs      =241400     tcpRetransBytes     =278717070
>>         tcpOutRsts          =223785     tcpOutFastRetrans   =   141
>>         tcpRttUpdate        =3633131    tcpTimRetrans       =136919
>>         tcpTimRetransDrop   =  3339     tcpTimKeepalive     = 33898
>>         tcpHalfOpenDrop     =     0     tcpOutSackRetrans   =186148
>>
>>


-- 


//////////////////////////////////////////////////////////////////
// Nathan Kroenert              nathan.kroenert at sun.com              //
// Senior Systems Engineer      Phone:  +61 3 9869 6255         //
// Global Systems Engineering   Fax:    +61 3 9869 6288         //
// Level 7, 476 St. Kilda Road                                  //
// Melbourne 3004   Victoria    Australia                       //
//////////////////////////////////////////////////////////////////

Reply via email to