Hello,

I've been digging through the archives, but I'm not seeing this issue.

On a machine that periodically crunches weblogs, I'm running ipfilter, and
I'm seeing what looks like the replies to the large amount of dns lookups
are periodically being blocked.  The box also acts as a dns server.
Currently I'm doing ipf on each host since no one here quite trusts
putting all the traffic through one pc (this will change eventually).

The relevant parts of the rules are (omitted martian blocks and explicit
passes for other hosts):

# allow the outside world these services
pass in quick on fxp0 proto tcp from any to any port = 25 flags S/SAFR
keep state keep frags
pass in quick on fxp0 proto tcp from any to any port = 53 flags S/SAFR
keep state keep frags
pass in quick on fxp0 proto udp from any to any port = 53 keep state
pass in quick on fxp0 proto tcp from any to any port = 80 flags S/SAFR
keep state keep frags
pass in quick on fxp0 proto tcp from any to any port = 443 flags S/SAFR
keep state keep frags

# set up state on fxp0 so that outgoing sessions can get a reply back

pass out quick on fxp0 proto icmp from any to any keep state keep frags
pass out quick on fxp0 proto udp from any to any keep state keep frags
pass out quick on fxp0 proto tcp from any to any keep state keep frags

Some examples of the blocked traffic (x.x.x.100 being the protected
machine):

Jan 21 19:53:23 box ipmon[67898]: 19:53:22.970830 fxp0 @0:30 b
207.155.248.5,53 -> x.x.x.100,17939 PR udp len 20 243 IN
Jan 21 19:53:23 box ipmon[67898]: 19:53:22.996375 fxp0 @0:30 b
12.109.100.2,53 -> x.x.x.100,18186 PR udp len 20 92 IN
Jan 21 19:53:23 box ipmon[67898]: 19:53:23.004858 fxp0 @0:30 b
207.155.248.5,53 -> x.x.x.100,14955 PR udp len 20 278 IN
Jan 21 19:53:23 box ipmon[67898]: 19:53:23.004888 fxp0 @0:30 b
207.155.252.5,53 -> x.x.x.100,8039 PR udp len 20 242 IN

I would think that the outgoing dns request would open up a return path
for any answers.  Since all the logs above show the blocked packets
originating from port 53, I assume these are answers to dns queries.  This
doesn't quite seem to be affecting performance, the full run on lookups
from the web log files happened as fast as it did before firing up ipf.

My only guesses are:

-some servers are behind some type of load balancer that's broken
-I'm overflowing something in the state table.  dnscache will fire up
 many requests at once.

Any ideas?  I'm just coming back to ipf after not touching it for a few
years.  I used the IPF how-to at obfuscation.org as a basis for my
ruleset.

Thanks,

Charles

--
Charles Sprickman
[EMAIL PROTECTED]


Reply via email to