Hi all,

after a lot of testing and changing everything (pmacct version, newer libpcap, 
kernel, Motherboard, switch) the reason behind the problem was quite trivial:
The network card (intel 1000e series) has some hardware acceleration functions 
that can assemble fragmented packets bevor showing them to the operating system…

The following commands disable these functions, and after that the results are 
reasonable again.
/sbin/ethtool -K eth0 tso off
/sbin/ethtool -K eth0 gso off
/sbin/ethtool -K eth0 gro off

cheers

Johannes

Am 26.02.2014 um 20:25 schrieb Johannes Formann <mlm...@formann.de>:

> Hi Paolo,
> 
> Thanks for your response.
> I tried the SVS-Version (is the repository on the website the right one, it 
> has 0.14 in the path) and have the same problems. Changing the buffer to the 
> size you suggested don’t help too.
> 
> Build it using this commands:
> mkdir pmacct-latest
> cd pmacct-latest/
> CVSROOT=:pserver:anonym...@cvs.pmacct.net:2402/home/repo-0.14
> export CVSROOT
> cvs login
> cvs co pmacct
> cd pmacct/
> ./configure --enable-l2=no --enable-ipv6 --enable-mysql
> make && make install
> 
> changed the /etc/init.d/pmacct the DAEMON-Line to use /usr/local/sbin/pmacctd
> service pmacct restart
> 
> I stumpled over an other bug too, the networks-File is ignored.
> 
> I think remote access should be possible but I have to ask someone else 
> before giving you login credentials.
> 
> cheers
> 
> Johannes
> 
> 
> Am 26.02.2014 um 19:04 schrieb Paolo Lucente <pa...@pmacct.net>:
> 
>> Hi Johannes,
>> 
>> This is indeed strange. I have two leads: there was a bug around
>> syslog logging being fixed recently (post 1.5.0rc2); the bug was
>> causing some memory corruption: can you, for a try, log directly
>> to a file via 'logfile' directive (or alternatively, slightly
>> more involved, upgrade to latesst CVS build). Should that not do
>> can you please try introducing some minimal internal buffering?
>> Something like the following should do:
>> 
>> plugin_pipe_size: 4096000
>> plugin_buffer_size: 40960
>> 
>> Let me know. Should none of the above work, would you be open
>> to some remote-access for a bit of debugging?
>> 
>> Cheers,
>> Paolo
>> 
>> On Tue, Feb 25, 2014 at 11:46:31PM +0100, Johannes Formann wrote:
>>> Hello,
>>> 
>>> I try to run pmacct on a debian 7 box.
>>> Basically everything runs fine, except sometimes pmacctd mud miscount the 
>>> traffic, since the average packetize goes over 1500 (latest size observed 
>>> about 7.000) wich is not possible (MTU-Limit of 1500 bytes).
>>> 
>>> Same problem with the debian packages an a fresh compiled version from 
>>> 1.5.rc2.
>>> 
>>> Any Ideas?
>>> 
>>> regards
>>> 
>>> Johannes
>>> 
>>> P.S. Used config 
>>> ! pmacctd configuration
>>> !
>>> !
>>> daemonize: true
>>> pidfile: /var/run/pmacctd.pid
>>> syslog: daemon
>>> aggregate[inbound]: dst_host
>>> aggregate[outbound]: src_host,proto
>>> aggregate[TMPflowSRC]: src_host,src_port,proto
>>> aggregate[TMPflowDST]: dst_host,proto
>>> aggregate_filter[TMPflowSRC]: udp
>>> aggregate_filter[TMPflowDST]: udp
>>> plugins: mysql[inbound], mysql[outbound], mysql[TMPflowSRC], 
>>> mysql[TMPflowDST]
>>> sql_table[inbound]: acct_%Y_%m_in
>>> sql_table[outbound]: acct_%Y_%m_out
>>> sql_table[TMPflowSRC]: acct_TMPflowSRC
>>> sql_table[TMPflowDST]: acct_TMPflowDST
>>> sql_table_schema[inbound]: /etc/pmacct/inbound.schema
>>> sql_table_schema[outbound]: /etc/pmacct/outbound.schema
>>> networks_file[inbound]: /etc/pmacct/networks
>>> networks_file[outbound]: /etc/pmacct/networks
>>> networks_file[TMPflowSRC]: /etc/pmacct/networks
>>> ports_file[TMPflowSRC]: /etc/pmacct/ports-udp 
>>> 
>>> ! on this interface
>>> interface: eth0
>>> 
>>> sql_db: pmacct
>>> sql_table_version: 4 
>>> sql_passwd: secret ;)
>>> sql_user: pmacct
>>> ! refresh the db every 1  minute
>>> sql_refresh_time: 60
>>> ! reduce the size of the insert/update clause
>>> sql_optimize_clauses: true
>>> ! accumulate values in each row for up to 5 minutes
>>> sql_history: 1m 
>>> ! create new rows on the minute, hour, day boundaries
>>> sql_history_roundoff: m
>>> ! Multivalue-Support, bei mysql schneller
>>> sql_multi_values: 12000000
>>> 
>>> ! viele Daten hier, mehr puffer
>>> sql_cache_entries: 64000
>>> 
>>> ports_file: /etc/pmacct/ports
>>> ! nur inserts
>>> sql_dont_try_update: true
>>> 
>>> ! intern puffern
>>> plugin_buffer_size: 163840
>>> plugin_pipe_size: 40960000
>>> 
>>> 
>>> _______________________________________________
>>> 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


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

Reply via email to