hello,
i was unable to find answers so i'm asking from here.
what does the following mean:
cflowd[2118]: [I] missed 107520 of 184440 flows from (null) engine 674104768 \
agg_method 0 (2.39511e-269% loss)
i get those from all of the routers, numbers differ each time, ofcourse.
and yes, i still get some data.
should i increase udp recv queue? it seems to be 232376 bytes. no, i
dont see non-zero values on 'netstat -a' recv-q.
also, earlier i also got:
cflowdmux[15388]: [E] semop(65536,0xbfbff128,1) failed to release buffer \
lock: Resource temporarily unavailable {CflowdPacketQueue.cc:438}
but it seemed to go away after i tweaked my configs and made shmem segment
16781312 instead of 8392704.
(yes, i get the first error with both shmem sizes.)
in case some more data is needed, i'm feeding flowdata from multiple
routers, and the amount is in many gigabytes per day. cflowd runs on
freebsd-current with "enough" of memory/cpu/disk/other.
configs:
** snip **
OPTIONS {
LOGFACILITY: local6
TCPCOLLECTPORT: 6667
PKTBUFSIZE: 16777216
TABLESOCKFILE: /var/run/cflowdtable.socket
FLOWDIR: /data/cflowd/flows
FLOWFILELEN: 4194304
NUMFLOWFILES: 2000
MINLOGMISSED: 100
}
COLLECTOR {
HOST: 127.0.0.1
ADDRESSES: { 127.0.0.1 }
AUTH: none
}
** snip **
from each router i collect:
{ protocol, portmatrix, ifmatrix, , nexthop, netmatrix, asmatrix, tos, flows }
** snip **
system {
logFacility: local6
dataDirectory: /log/cflowd
filePrefix: arts
pidFile: /var/run/cfdcollect.pid
}
cflowd {
host: localhost
tcpCollectPort: 6667
minPollInterval: 180
}
** snip **
mickey
-- cflowd mailing list cflowd@caida.org
This archive was generated by hypermail 2b29 : Fri Jul 06 2001 - 00:20:34 PDT