Unusual label behavior

Craig Merchant cmerchant at responsys.com
Tue Jul 23 22:30:16 EDT 2013


Ralabel.conf:

RALABEL_IANA_ADDRESS=yes
RALABEL_IANA_ADDRESS_FILE="/usr/local/argus/responsys-iana-file"
#RALABEL_IANA_ADDRESS_FILE="/usr/local/argus/iana-address-file.full"
RALABEL_ARIN_COUNTRY_CODES=yes
#RALABEL_BIND_NAME="all"
RALABEL_IANA_PORT=no
#RALABEL_IANA_PORT_FILE="/usr/local/argus/iana-port-numbers"
RALABEL_ARGUS_FLOW=no
RALABEL_ARGUS_FLOW_FILE="/usr/local/argus/label-file"
#RALABEL_GEOIP_ASN=yes
#RALABEL_GEOIP_ASN_FILE="/usr/local/share/GeoIP/GeoIPASNum.dat"
#       RALABEL_GEOIP_CITY="saddr,daddr:lat/lon"
#       RALABEL_GEOIP_CITY="*:city,region,cname,lat,lon"
RALABEL_GEOIP_CITY="saddr,daddr,inode:cco,cname"
RALABEL_GEOIP_CITY_FILE="/usr/local/argus/GeoIPCity.dat"

I grab my data with rastream and  then use the scripting feature to do some post-processing:

/usr/local/bin/rastream -S 10.10.10.10:561 -M time 5m -B 10s -w /ssd/argus/%s.argus -f /usr/local/argus/rastream.sh &

Rastream.sh looks like:

racluster -r $FILES -M correct -m proto saddr daddr dport -c "@" -p 3 -u -Z b -n –s "+0ltime,+trans,+dur,+runtime,+mean,+stddev,+sum,+sco,+dco,+spkts                 ,+dpkts,+sbytes,+dbytes,+load,+sload,+dload,+loss,+sloss,+dloss,+ploss,+sploss,+dploss,+rate,+srate,+drate,+appbytes,+sappbytes,+dappbytes,+label:200,+abr" - src net 10.0.0.0/8 and dst net 10.0.0.0/8 | sed -e 's/,/#/g' | sed -e 's/@/,/g' > /ssd/argus/splunk/internal.csv

I found that variable length label fields that were comma separated confused Splunk when I imported the CSV file.  So, I run it through sed to change the label field’s commas to # and then change the @ delimiter back to a comma.

Thx.

Craig

From: Carter Bullard [mailto:carter at qosient.com]
Sent: Tuesday, July 23, 2013 7:17 PM
To: Craig Merchant
Cc: Argus (argus-info at lists.andrew.cmu.edu)
Subject: Re: [ARGUS] Unusual label behavior

Labels are strings that are assembled in big global buffers, and we need to work on concurrent processing, especially when the label configurations have "cont", where multiple appended strings are applied to the same flow, which can push us  to overflow, which can cause string collision, which looks to be the problem your reporting.

So need to understand how the processing goes.  argus -> radium(labeler) -> ra ?

So need to see your ralabel.conf file to see what could be wrong.

Carter

On Jul 23, 2013, at 5:24 PM, Craig Merchant <cmerchant at responsys.com<mailto:cmerchant at responsys.com>> wrote:
Hey, Carter…

Since we upgraded to the latest version, I’m getting some really strange behavior with labels.  Some events have massive incorrect labels.  Other events that should have labels don’t.

For example, one event had the following in the label field:

saddr=internal,DC2,I5-PSO,app,dev-tserv2,app,dev-tserv2,app,qa-tserv2,app,qa-tserv2,app,ri5-tserv2,app,ri5-tserv2,app,ri5-tserv2,app,ri5-tserv2,app,ri5-tserv2,app,ri5-tserv2,app,sp1-tserv2,app,stage3-tserv2,app,stage3-tserv2,etldev,tserv2,Cpn01:tserv2=etlserv,tserv2,etlserv,tserv2,sa,tserv2,sa,tserv2,sa,tserv2,sa,tserv2,sa,tserv2,database,database,database,database,database,database,database,database,DME,DC2,ETL,Dev,DME,DC2,ETL,Dev,DME,DB:daddr=internal,DC2,SysOps-Admin,bi-etl,dc2,dbadmin,dc2,dbmonitor,dc2,dns-master,dc2,etltest,dc2,fly,uber,fly,uber,ldap-master,dc2,ldap-master,dc2,mailscan,uber,mon,dc2,mondb,dc2,mta,uber,mta,uber,prodops:dc2=prodopsdb,dc2,rac,01n01-dc2,rac,01n02-dc2,rac,02n01-dc2,rac,02n02-dc2,sa,dc2,sa,dc2,sa,dc2,zabbix,uber,zabbixdb,uber,database,database,database,database,database,database"

The correct label should have been:

saddr=internal,DC2,I5-PSO,app,ri5-tserv2:daddr=internal,DC2,SysOps-Admin,fly,uber

I checked through the label file for some of the hosts that were missing labels and there were definitely records that should have had labels that didn’t.

If you still have that tcpdump I sent you, I can forward you the latest version of the label file and see if you can replicate the results.

Thanks.

Craig
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist1.pair.net/pipermail/argus/attachments/20130724/9f319ee8/attachment.html>


More information about the argus mailing list