Hi Alex,

Thanks for your feedback. I see you did run "tcpdump -n -vv -i nflog:1"
which is equivalent to run uacctd without any filters; as you may know,
you can append a BPF-style filter to the tcpdump command-line, precisely
as you express it in pre_tag_map. Can you give that a try and see if you
get any luck?

My expextation is: if something does not work with pre_tag_map, it
should also not work with tcpdump; if you work out a filter to work
against tcpdump, that should work in pre_tag_map as well. Any disconnect
among the two may bring the scent of a bug.

Paolo
 
On Tue, Feb 25, 2020 at 11:20:21AM +0200, Alex K wrote:
> Here is the output when running in debug mode:
> 
> INFO ( default/core ): Linux NetFilter NFLOG Accounting Daemon, uacctd
> (20200222-01)
> INFO ( default/core ):  '--prefix=/usr' '--enable-mysql' '--enable-nflog'
> '--enable-l2' '--enable-traffic-bins' '--enable-bgp-bins'
> '--enable-bmp-bins' '--enable-st-bins'
> INFO ( default/core ): Reading configuration file
> '/root/pmacct/uacctd2.conf'.
> INFO ( print_wan0_in/print ): plugin_pipe_size=4096000 bytes
> plugin_buffer_size=280 bytes
> INFO ( print_wan0_in/print ): ctrl channel: obtained=212992 bytes
> target=117024 bytes
> INFO ( print_wan0_out/print ): plugin_pipe_size=4096000 bytes
> plugin_buffer_size=280 bytes
> INFO ( print_wan0_out/print ): ctrl channel: obtained=212992 bytes
> target=117024 bytes
> INFO ( print_wan0_in/print ): cache entries=16411 base cache
> memory=54878384 bytes
> INFO ( default/core ): [pretag2.map] (re)loading map.
> INFO ( print_wan0_out/print ): cache entries=16411 base cache
> memory=54878384 bytes
> INFO ( default/core ): [pretag2.map] map successfully (re)loaded.
> INFO ( default/core ): [pretag2.map] (re)loading map.
> INFO ( default/core ): [pretag2.map] map successfully (re)loaded.
> INFO ( default/core ): [pretag2.map] (re)loading map.
> INFO ( default/core ): [pretag2.map] map successfully (re)loaded.
> INFO ( default/core ): Successfully connected Netlink NFLOG socket
> 
> It doesn't seem to have any issues loading the maps, though it is not
> collecting anything. When capturing with tcpdump I see packets going
> through:
> 
> tcpdump -n -vv -i nflog:1
> 09:16:05.831131 IP (tos 0x0, ttl 64, id 36511, offset 0, flags [DF], proto
> ICMP (1), length 84)
>     192.168.28.11 > 8.8.8.8: ICMP echo request, id 17353, seq 1, length 64
> 09:16:05.831362 IP (tos 0x0, ttl 49, id 0, offset 0, flags [none], proto
> ICMP (1), length 84)
>     8.8.8.8 > 192.168.28.11: ICMP echo reply, id 17353, seq 1, length 64
> 09:16:05.831392 IP (tos 0x0, ttl 64, id 36682, offset 0, flags [DF], proto
> ICMP (1), length 84)
>     192.168.28.11 > 8.8.8.8: ICMP echo request, id 17353, seq 2, length 64
> 09:16:06.855200 IP (tos 0x0, ttl 49, id 0, offset 0, flags [none], proto
> ICMP (1), length 84)
>     8.8.8.8 > 192.168.28.11: ICMP echo reply, id 17353, seq 2, length 64
> 
> The pmacct  version I am running is latest master.
> Thank you for your assistance.
> 
> Alex
> 
> 
> On Mon, Feb 24, 2020 at 6:20 PM Alex K <rightkickt...@gmail.com> wrote:
> 
> > Hi Paolo,
> >
> > On Sat, Feb 22, 2020 at 4:18 PM Paolo Lucente <pa...@pmacct.net> wrote:
> >
> >>
> >> Hi Alex,
> >>
> >> Is it possible with the new setup - the one where pre_tag_map does not
> >> match anything - the traffic is VLAN-tagged (or MPLS-labelled)? If so,
> >> you should adjust filters accordingly and add 'vlan and', ie. "vlan and
> >> src net 192.168.28.0/24 or vlan and src net 192.168.100.0/24".
> >>
> > The traffic is not VLAN or MPLS. It is simple one. I confirm I can collect
> > traffic when removing the pretag directives. Also when stopping uacctd, I
> > can capture traffic at nflog:1 interface.
> > I simplified the configuration as below:
> >
> > !
> > daemonize: true
> > promisc:   false
> > uacctd_group: 1
> > !
> > pre_tag_map: pretag2.map
> > pre_tag_filter[print_wan0_in]: 1
> > pre_tag_filter[print_wan0_out]: 2
> > !
> > !-------------------------------------------------
> > plugins: print[print_wan0_in], print[print_wan0_out]
> > print_refresh_time: 10
> > print_history: 15m
> > print_output_file_append: true
> > !
> > print_output[print_wan0_in]: csv
> > print_output[print_wan0_out]: csv
> > print_output_file[print_wan0_in]: traffic-wan0-in.csv
> > print_output_file[print_wan0_out]: traffic-wan0-out.csv
> > !
> > aggregate[print_wan0_in]: tag, src_host, dst_host, src_port, dst_port,
> > proto
> > aggregate[print_wan0_out]: tag, src_host, dst_host, src_port, dst_port,
> > proto
> > !
> >
> > with pretag2.map
> > set_tag=1 filter='src net 192.168.28.0/24'
> > set_tag=2 filter='dst net 192.168.28.0/24'
> >
> > As soon as I enable the pretag directives as below, I do not see any
> > traffic being collected from uacctd at NFLOG goup 1
> >
> > pre_tag_map: pretag2.map
> > pre_tag_filter[print_wan0_in]: 1
> > pre_tag_filter[print_wan0_out]: 2
> >
> > I am running pmacct 1.7.4.
> >
> >
> >> Paolo
> >>
> >> On Fri, Feb 21, 2020 at 01:04:25PM +0200, Alex K wrote:
> >> > Working further on this, it seems that for pmacct is sufficient to
> >> filter
> >> > traffic using only the pre_tag_filter, thus no need for the aggregation
> >> > filters.
> >> > The issue with this setup though is that I loose the information of the
> >> > pre_nat source IP address when monitoring at the WAN interfaces. Due to
> >> > this I am switching to uacctd as following:
> >> >
> >> > !
> >> > daemonize: true
> >> > promisc:   false
> >> > uacctd_group: 1
> >> > !networks_file: networks.lst
> >> > !ports_file: ports.lst
> >> > !
> >> > pre_tag_map: pretag2.map
> >> > pre_tag_filter[print_wan0_in]: 1
> >> > pre_tag_filter[print_wan0_out]: 2
> >> > pre_tag_filter[wan0_in]: 1
> >> > pre_tag_filter[wan0_out]: 2
> >> > !
> >> > plugins: print[print_wan0_in], print[print_wan0_out], mysql[wan0_in],
> >> > mysql[wan0_out]
> >> > plugin_pipe_size[wan0_in]: 1024000
> >> > plugin_pipe_size[wan0_out]: 1024000
> >> > print_refresh_time: 10
> >> > print_history: 15m
> >> > print_output_file_append: true
> >> > !
> >> > print_output[print_wan0_in]: csv
> >> > print_output_file[print_wan0_in]: in_traffic.csv
> >> > print_output[print_wan0_out]: csv
> >> > print_output_file[print_wan0_out]: out_traffic.csv
> >> > !
> >> > aggregate[print_wan0_in]: dst_host, src_port, dst_port, proto
> >> > aggregate[print_wan0_out]: src_host, src_port, dst_port, proto
> >> > !
> >> > sql_table[wan0_in]: traffic_wan0_in_%Y%m%d_%H%M
> >> > sql_table[wan0_out]: traffic_wan0_out_%Y%m%d_%H%M
> >> > !
> >> > sql_table_schema[wan0_in]: traffic_wan0_in.schema
> >> > sql_table_schema[wan0_out]: traffic_wan0_out.schema
> >> > !
> >> > sql_host: localhost
> >> > sql_db : uacct
> >> > sql_user : uacct
> >> > sql_passwd: uacct
> >> > sql_refresh_time: 30
> >> > sql_optimize_clauses: true
> >> > sql_history : 24h
> >> > sql_history_roundoff: mhd
> >> > !
> >> > aggregate[wan0_in]: dst_host, src_port, dst_port, proto
> >> > aggregate[wan0_out]: src_host, src_port, dst_port, proto
> >> >
> >> > Where pretag2.map:
> >> > set_tag=1 filter='src net 192.168.28.0/24 or src net 192.168.100.0/24'
> >> > set_tag=2 filter='dst net 192.168.28.0/24 or dst net 192.168.100.0/24'
> >> >
> >> > The issue I have with the above config is that no traffic is being
> >> > collected at all. I confirm that when removing the pre_tag filters,
> >> traffic
> >> > is collected, though it is not sorted per direction as I would like to
> >> > have.
> >> > Can I use pre_tag_map and pre_tag_filter with uacctd? I don't see any
> >> > examples for uacctd at
> >> >
> >> https://github.com/pmacct/pmacct/blob/master/examples/pretag.map.example.
> >> >
> >> > Thanx,
> >> > Alex
> >> >
> >> > On Thu, Feb 20, 2020 at 6:33 PM Alex K <rightkickt...@gmail.com> wrote:
> >> >
> >> > > Hi all,
> >> > >
> >> > > I have a router with multiple interfaces and will need to account
> >> traffic
> >> > > at its several WAN interfaces. My purpose is toaccount the traffic
> >> with the
> >> > > tuple details and the direction.
> >> > >
> >> > > As a test I have compiled the following simple configuration for
> >> pmacctd:
> >> > >
> >> > > !
> >> > > daemonize: true
> >> > > plugins: print[wan0_in], print[wan0_out]
> >> > > print_refresh_time: 10
> >> > > print_history: 15m
> >> > > !
> >> > > print_output[wan0_in]: csv
> >> > > print_output_file[wan0_in]: in_traffic.csv
> >> > > print_output[wan0_out]: csv
> >> > > print_output_file[wan0_out]: out_traffic.csv
> >> > > !
> >> > > aggregate[wan0_in]: src_host, dst_host, src_port, dst_port, tag
> >> > > aggregate[wan0_out]: src_host, dst_host, src_port, dst_port, tag
> >> > > !
> >> > > pre_tag_filter[wan0_in]:1
> >> > > pre_tag_filter[wan0_out]:2
> >> > > !
> >> > > pcap_interface: eth0
> >> > > pre_tag_map: pretag.map
> >> > > networks_file: networks.lst
> >> > > ports_file: ports.lst
> >> > > !
> >> > >
> >> > > where pretag.map is:
> >> > > set_tag=1 filter='ether dst 52:54:00:69:a6:0b'
> >> > > set_tag=2 filter='ether src 52:54:00:69:a6:0b'
> >> > >
> >> > > and networks.lst is:
> >> > > 10.100.100.0/24
> >> > >
> >> > > It seems that the details output at the CSV are correctly filtered
> >> > > according to the tag, thus recording the direction also, based on the
> >> MAC
> >> > > address of the WAN0 interface.
> >> > >
> >> > > Is this the correct approach to achieve this or is there any other
> >> > > recommended way? Do I need to use aggregate_filters?
> >> > >
> >> > > Also, although I have set a network filter to capture only
> >> 10.100.100.0/24,
> >> > > I observe several networks in/out being collected, indicating that the
> >> > > network_file directive is ignored or I have misunderstood its
> >> purpose. My
> >> > > purpose it to collect traffic only generated from subnets that belong
> >> to
> >> > > configured interfaces of the router.
> >> > >
> >> > > Thanx for your feedback!
> >> > > Alex
> >> > >
> >> > >
> >> > >
> >>
> >> > _______________________________________________
> >> > pmacct-discussion mailing list
> >> > http://www.pmacct.net/#mailinglists
> >>
> >>
> >> _______________________________________________
> >> pmacct-discussion mailing list
> >> http://www.pmacct.net/#mailinglists
> >>
> >

_______________________________________________
pmacct-discussion mailing list
http://www.pmacct.net/#mailinglists

Reply via email to