Hi,
I'm seeing fairly high values for 'failed connection attempts' reported
by netstat -s. The values reported just under 10% of the 'active
connections openings' value. The full output from netstat -s is below.
Although we've not noticed any real problems I was just wondering what
causes the failed connection attempts, are they indicative of a some
resource shortage with the server, an indication of miss-behaving
clients or network hardware or just a fact of life?
The machine is purely an SSL webserver running apache 1.3.28 on RedHat
AS2.1 (2.4.9-e.3enterprise). The box is a 4-way 2GHz P4 Xeon with HT
enabled (so 8 virtual cpus). The network cards are reported as "BROADCOM
Corporation: Unknown device 16a7 (rev 02)" and are using the tg3 driver.
Both cards are connected into switches and running at 100Mb FD, ifconfig
has not reported any dropped packets, errors, overruns etc.
I have not seen the server go below 50% idle during busy periods and it
has plenty of free memory.
Grateful for any insight, thanks
Will
output from netstat -s
Ip:
1133111899 total packets received
0 forwarded
0 incoming packets discarded
1131645930 incoming packets delivered
1037009075 requests sent out
Icmp:
2940 ICMP messages received
1 input ICMP message failed.
ICMP input histogram:
destination unreachable: 2937
echo requests: 3
2991 ICMP messages sent
0 ICMP messages failed
ICMP output histogram:
destination unreachable: 2988
echo replies: 3
Tcp:
18484519 active connections openings
0 passive connection openings
1366079 failed connection attempts
0 connection resets received
7 connections established
1133078991 segments received
1037003138 segments send out
5322606 segments retransmited
23640 bad segments received.
73914 resets sent
Udp:
10 packets received
2936 packets to unknown port received.
0 packet receive errors
2946 packets sent
TcpExt:
4798 resets received for embryonic SYN_RECV sockets
ArpFilter: 0
86855272 TCP sockets finished time wait in fast timer
14833625 delayed acks sent
9859 delayed acks further delayed because of locked socket
Quick ack mode was activated 86602 times
665622 times the listen queue of a socket overflowed
665622 SYNs to LISTEN sockets ignored
494190272 packets directly queued to recvmsg prequeue.
1589965710 packets directly received from backlog
2204994422 packets directly received from prequeue
2102458 packets header predicted
257644033 packets header predicted and directly queued to user
TCPPureAcks: 274746597
TCPHPAcks: 163586921
TCPRenoRecovery: 776880
TCPSackRecovery: 4833
TCPSACKReneging: 2
TCPFACKReorder: 1
TCPSACKReorder: 0
TCPRenoReorder: 54706
TCPTSReorder: 3
TCPFullUndo: 33
TCPPartialUndo: 11
TCPDSACKUndo: 0
TCPLossUndo: 15
TCPLoss: 2786
TCPLostRetransmit: 0
TCPRenoFailures: 253385
TCPSackFailures: 4299
TCPLossFailures: 147036
TCPFastRetrans: 1050073
TCPForwardRetrans: 164
TCPSlowStartRetrans: 850324
TCPTimeouts: 2260969
TCPRenoRecoveryFail: 173604
TCPSackRecoveryFail: 925
TCPSchedulerFailed: 9062
TCPRcvCollapsed: 0
TCPDSACKOldSent: 1307
TCPDSACKOfoSent: 0
TCPDSACKRecv: 225
TCPDSACKOfoRecv: 0
TCPAbortOnSyn: 62
TCPAbortOnData: 1309
TCPAbortOnClose: 174
TCPAbortOnMemory: 0
TCPAbortOnTimeout: 37892
TCPAbortOnLinger: 0
TCPAbortFailed: 0
TCPMemoryPressures: 0
--
Will Slater Will.Slater@orbisuk.com
Orbis Tel: +44 (0)20 8742 1600
www.orbisuk.com Fax: +44 (0)20 8742 2649
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
