[lug] Dropped packet question

Chip Atkinson chip at pupman.com
Thu Sep 26 10:25:38 MDT 2013


Wow.  I've never used that.  I'll paste in below what it shows.  I see
things like outgoing packets dropped, reassemblies, fragments, etc.
What should I be looking for in particular?
Oddly too, it looks that eth1 and eth0 have the same numbers or very close
to the same (since it's actively sending and receiving traffic)

chip1:/var/log # netstat -s eth0
Ip:
    18528785 total packets received
    61576 forwarded
    0 incoming packets discarded
    18229593 incoming packets delivered
    18372091 requests sent out
    560 outgoing packets dropped
    47913643 reassemblies required
    15979146 packets reassembled ok
    31957472 fragments received ok
    47912823 fragments created
Icmp:
    82067 ICMP messages received
    126 input ICMP message failed.
    ICMP input histogram:
        destination unreachable: 24598
        timeout in transit: 1167
        echo requests: 37000
        echo replies: 19302
    45902 ICMP messages sent
    0 ICMP messages failed
    ICMP output histogram:
        destination unreachable: 8902
        echo replies: 37000
Tcp:
    90757 active connections openings
    74760 passive connection openings
    631 failed connection attempts
    1725 connection resets received
    6 connections established
    1488263 segments received
    1565368 segments send out
    20421 segments retransmited
    9 bad segments received.
    100442 resets sent
Udp:
    16653309 packets received
    5151 packets to unknown port received.
    0 packet receive errors
    16680883 packets sent
TcpExt:
    145 invalid SYN cookies received
    29 resets received for embryonic SYN_RECV sockets
    ArpFilter: 0
    28479 TCP sockets finished time wait in fast timer
    142 packets rejects in established connections because of timestamp
    25313 delayed acks sent
    61 delayed acks further delayed because of locked socket
    Quick ack mode was activated 1483 times
    54672 packets directly queued to recvmsg prequeue.
    366613 packets directly received from backlog
    1968664 packets directly received from prequeue
    178056 packets header predicted
    40329 packets header predicted and directly queued to user
    TCPPureAcks: 379540
    TCPHPAcks: 451987
    TCPRenoRecovery: 5
    TCPSackRecovery: 897
    TCPSACKReneging: 7
    TCPFACKReorder: 58
    TCPSACKReorder: 0
    TCPRenoReorder: 3
    TCPTSReorder: 48
    TCPFullUndo: 91
    TCPPartialUndo: 499
    TCPDSACKUndo: 51
    TCPLossUndo: 1027
    TCPLoss: 552
    TCPLostRetransmit: 0
    TCPRenoFailures: 0
    TCPSackFailures: 439
    TCPLossFailures: 43
    TCPFastRetrans: 1600
    TCPForwardRetrans: 284
    TCPSlowStartRetrans: 398
    TCPTimeouts: 6235
    TCPRenoRecoveryFail: 0
    TCPSackRecoveryFail: 155
    TCPSchedulerFailed: 0
    TCPRcvCollapsed: 0
    TCPDSACKOldSent: 1472
    TCPDSACKOfoSent: 0
    TCPDSACKRecv: 2102
    TCPDSACKOfoRecv: 5
    TCPAbortOnSyn: 0
    TCPAbortOnData: 3261
    TCPAbortOnClose: 1081
    TCPAbortOnMemory: 0
    TCPAbortOnTimeout: 950
    TCPAbortOnLinger: 0
    TCPAbortFailed: 0
    TCPMemoryPressures: 0


On Fri, 27 Sep 2013, David Frye wrote:

> What does 'netstat -s <interface>' show?
> 
> If packets are being dropped, it may have additional information there not shown with ifconfig.
> 
> 
> On Sep 26, 2013, at 8:53 AM, Chip Atkinson wrote:
> 
> > Thanks.  I'm not seeing errors or dropped packets in ifconfig, which is
> > kind of weird, isn't  it?  If ping reports dropped packets, wouldn't that
> > droppage appear in the output of ifconfig?  
> > 
> > On Fri, 27 Sep 2013, Dan Ferris wrote:
> > 
> >> Start with something easy.  Check ifconfig and see if there are errors 
> >> on the interface.  If so, then start by checking hardware. You could 
> >> have a bad cable, bad nic, bad switch port, or a duplex mismatch.
> >> 
> >> Dan
> >> 
> >> On 9/27/2013 9:31 AM, Davide Del Vento wrote:
> >>> Since you control the server, don't the logs tell you something about
> >>> the dropped packets? Since you don't see drops with the netbook, you
> >>> can rule out the rest of the network: it must be the server box.
> >>> 
> >>> It may be dropping packets for a variety of reasons, just to mention a
> >>> couple of stupid ones: a defective network card or too high CPU load.
> >>> 
> >>> Cheers,
> >>> Davide
> >>> 
> >>> On Thu, Sep 26, 2013 at 8:48 AM, Chip Atkinson <chip at pupman.com> wrote:
> >>>> Greetings all,
> >>>> 
> >>>> Due to the recent flooding I had to change data centers from my parents'
> >>>> basement to mine, which resulted in re-doing my network.
> >>>> 
> >>>> Now that I've moved and re-IPed the server, I'm seeing large numbers of
> >>>> dropped packets, slow ping times, basic network malaise.  I've been
> >>>> running a series of 100 pings 5 sec apart and then looking at the reported
> >>>> loss figures.
> >>>> 
> >>>> With comcast's help, I believe that we've eliminated them and their
> >>>> hardware.
> >>>> 
> >>>> I put a small linux netbook on the network in place of the server and was
> >>>> able to ping it from outside (vpn to work and out from there) and the
> >>>> ping response time and dropped packets were basically gone.  Besides being
> >>>> newer hardware and OS, the netbook had no services (web, dns, email).
> >>>> 
> >>>> I then connected the server and see the dropped packet and slow ping time
> >>>> issue again.
> >>>> 
> >>>> I was using tcpdump and noticed that a large portion of the traffic is DNS
> >>>> lookups:
> >>>> 
> >>>> 08:42:23.411809 IP (tos 0x0, ttl  64, id 42252, offset 0, flags [+],
> >>>> length: 1500) 173.14.7.2.53 > 108.174.149.7.2305:  13490| 250/0/1
> >>>> bitstress.com. SOA[|domain]
> >>>> 08:42:23.411817 IP (tos 0x0, ttl  64, id 42252, offset 1480, flags [+],
> >>>> length: 1500) 173.14.7.2 > 108.174.149.7: udp
> >>>> 08:42:23.411822 IP (tos 0x0, ttl  64, id 42252, offset 2960, flags [none],
> >>>> length: 1150) 173.14.7.2 > 108.174.149.7: udp
> >>>> 
> >>>> Googling found this:
> >>>> http://dnsamplificationattacks.blogspot.com/2013/09/domain-bitstresscom.html
> >>>> 
> >>>> My question is whether or not the dns traffic could be responsible for all
> >>>> the dropped network packets or should I start looking elsewhere for the
> >>>> problem?
> >>>> 
> >>>> I switched network interfaces and took the original server network
> >>>> interface off the network, thinking that it could be broadcasting a bunch
> >>>> of noise but still am seeing packet losses, though perhaps not as severe.
> >>>> 
> >>>> 
> >>>> Thanks in advance for any insight and help.
> >>>> 
> >>>> Chip
> >>>> 
> >>>> 
> >>>> _______________________________________________
> >>>> Web Page:  http://lug.boulder.co.us
> >>>> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> >>>> Join us on IRC: irc.hackingsociety.org port=6667 channel=#hackingsociety
> >>> _______________________________________________
> >>> Web Page:  http://lug.boulder.co.us
> >>> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> >>> Join us on IRC: irc.hackingsociety.org port=6667 channel=#hackingsociety
> >> 
> >> _______________________________________________
> >> Web Page:  http://lug.boulder.co.us
> >> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> >> Join us on IRC: irc.hackingsociety.org port=6667 channel=#hackingsociety
> >> 
> > 
> > _______________________________________________
> > Web Page:  http://lug.boulder.co.us
> > Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> > Join us on IRC: irc.hackingsociety.org port=6667 channel=#hackingsociety
> 
> 
> --
> David
> dafr at dafr.us
> 
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> Join us on IRC: irc.hackingsociety.org port=6667 channel=#hackingsociety
> 



More information about the LUG mailing list