Very Strange - TCP SWEEP Alerts / Inconsistent with traffic on system

khatfield at khatfield at
Sun Jun 27 21:22:51 UTC 2010

 We have a strange situation occurring lately where we are getting some reports of TCP Sweeps from some one of our IP's, yet the IP is one of many specifically configured for inbound traffic and do not emit outbound traffic unless for response. Specifically, these are ddos mitigation IP's so they are attacked fairly frequently. With this in mind, the last few days one of the IP's being reported has been under constant attack.

Here is an example report we received from AT&T:
04:29:27 x.x.x.x [TCP-SWEEP] (total=23,dp=1024,min=,max=,Jun27-04:21:01,Jun27-04:29:26) (USI-amsxaid01)
04:29:27 x.x.x.x [TCP-SWEEP] (total=16,dp=3072,min=,max=,Jun27-04:21:15,Jun27-04:29:09) (USI-amsxaid01)
04:36:44 x.x.x.x [TCP-SWEEP] (total=16,dp=1024,min=,max=,Jun27-04:29:51,Jun27-04:35:53) (USI-amsxaid01)
04:20:47 x.x.x.x [TCP-SWEEP] (total=25,dp=1024,min=,max=,Jun27-04:12:37,Jun27-04:20:40) (USI-amsxaid01)
04:20:47 x.x.x.x [TCP-SWEEP] (total=18,dp=3072,min=,max=,Jun27-04:13:15,Jun27-04:20:37) (USI-amsxaid01)
04:12:36 x.x.x.x [TCP-SWEEP] (total=34,dp=1024,min=,max=,Jun27-03:56:28,Jun27-04:12:29) (USI-amsxaid01)
04:12:36 x.x.x.x [TCP-SWEEP] (total=28,dp=3072,min=,max=,Jun27-03:56:48,Jun27-04:11:45) (USI-amsxaid01)
Report from DK*CERT:
If nothing else mentioned below, timezone is believed to be UTC+0200(CEST)
Destination address(es): Adresser i nettene og

Security logs:
#Jun 27 18:13:40 2010 .. Jun 27 18:58:13 2010
# Scan from x.x.x.x affecting at least
# 81 addresses targeting TCP:1024, TCP:3072.
I have removed our IP and replaced it with x.x.x.x.  To be a bit more clear, this is a reverse-proxy IP address. This IP is in a NAT type configuration where it is sent back to filtering clusters. No outbound traffic is configured on these IP's except where requests / responses flow through it.

I know a year or two ago there was a bug in Cisco IOS that would report a sweep when extreme packet load occurred or a burst hit. At the time of this report we saw an attack burst to around 310,000PPS on this IP (inbound). Is it simply likely the networks reporting have several IP's being used in the attack and that is what they are seeing? That's what we originally thought but the port scans throw that theory off... Our security team has gone through all PCAPs during the mentioned time frames and we are not showing any sort of outbound scan traffic.

Any ideas why this would be showing as a sweep? Our IDS systems do not scan requesting IP's originating systems. Any help is appreciated, we're simply trying to get to the bottom of the reports.


More information about the NANOG mailing list