You need to sign in or sign up before continuing.
netrec metrics
Created by: DrQz
KDR netrec fields currently look like this:
> redrooz@earth:~$ /opt/kronometrix/bin/netrec -h
> USAGE: netrec [-hV] | [interval [count]]
> e.g. netrec 5 print continuously, every 5 seconds
> netrec 1 5 print 5 times, every 1 second
>
> FIELDS:
> #01 timestamp : seconds since Epoch, time
> #02 active : TCP active connections, number
> #03 passive : TCP passive connections, number
> #04 failed : TCP failed connection attempts, number
> #05 recv_rst : TCP connection resets received, number
> #06 est : TCP connections established, number
> #07 sent_rst : TCP resets sent, number
> #08 upks_recv : UDP packets received, number
> #09 upks_sent : UDP packets sent, number
> #10 fast_retr : fast retransmits, number
> #11 fwd_retr : forward retransmits, number
> #12 slow_retr : retransmits in slow start, number
This seems woefully insufficient. It should contain snd/rcv and errors from all the ISO levels available in the netstat -s command (ICMP, TCP, IP, etc.), at least:
redrooz@earth:~$ netstat -s
Ip:
1017388057 total packets received
0 forwarded
0 incoming packets discarded
1016756023 incoming packets delivered
906246357 requests sent out
4 dropped because of missing route
Icmp:
1101182 ICMP messages received
659691 input ICMP message failed.
ICMP input histogram:
destination unreachable: 1101172
echo replies: 10
162 ICMP messages sent
0 ICMP messages failed
ICMP output histogram:
destination unreachable: 42
echo request: 120
IcmpMsg:
InType0: 10
InType3: 1101172
OutType3: 42
OutType8: 120
Tcp:
3637262 active connections openings
318 passive connection openings
1800039 failed connection attempts
45 connection resets received
2 connections established
1013614523 segments received
1477898340 segments send out
964450 segments retransmited
1 bad segments received.
1500837 resets sent
Udp:
1918494 packets received
38 packets to unknown port received.
0 packet receive errors
1918614 packets sent
UdpLite:
TcpExt:
397 invalid SYN cookies received
9 ICMP packets dropped because they were out-of-window
68461 TCP sockets finished time wait in fast timer
3838534 delayed acks sent
347 delayed acks further delayed because of locked socket
Quick ack mode was activated 7184 times
4811310 packets directly queued to recvmsg prequeue.
4114119174 bytes directly in process context from backlog
3268755522 bytes directly received in process context from prequeue
447511935 packet headers predicted
9800132 packets header predicted and directly queued to user
16862289 acknowledgments not containing data payload received
706174761 predicted acknowledgments
16637 times recovered from packet loss by selective acknowledgements
18 bad SACK blocks received
Detected reordering 38 times using SACK
Detected reordering 17 times using time stamp
19 congestion windows fully recovered without slow start
65 congestion windows partially recovered using Hoe heuristic
3368 congestion windows recovered without slow start by DSACK
61 congestion windows recovered without slow start after partial ack
TCPLostRetransmit: 492
3 timeouts after SACK recovery
1 timeouts in loss state
89956 fast retransmits
5777 forward retransmits
557 retransmits in slow start
435680 other TCP timeouts
TCPLossProbes: 406976
TCPLossProbeRecovery: 245887
21 SACK retransmits failed
7194 DSACKs sent for old packets
398332 DSACKs received
36 DSACKs for out of order packets received
10 connections reset due to unexpected data
35 connections reset due to early user close
1 connections aborted due to timeout
TCPSACKDiscard: 9
TCPDSACKIgnoredOld: 8
TCPDSACKIgnoredNoUndo: 192079
TCPSpuriousRTOs: 13
TCPSackShifted: 206249
TCPSackMerged: 89105
TCPSackShiftFallback: 172606
TCPRetransFail: 159
TCPRcvCoalesce: 16536810
TCPOFOQueue: 460634
TCPChallengeACK: 7
TCPSYNChallenge: 1
TCPSpuriousRtxHostQueues: 2250
IpExt:
InBcastPkts: 121784
InOctets: 854093680
OutOctets: -162930218
InBcastOctets: 36094302
InNoECTPkts: 1109844394
Moreover, these are just counts. It is generally more helpful is they are represented as throughput metrics (i.e, counts per sample interval) because that is more closely allied with the concept of network bandwidth.
Edited by Stefan Parvu