I'm using MySQL 4.1.14 Standard.  This is the default v2 table schema.  
I will try to downgrade and see if it makes any difference.

Here is the full config:

--begin--
debug:                false
daemonize:            false

plugins:                    mysql[inbound], mysql[outbound]
aggregate[inbound]:         tag,dst_host
aggregate[outbound]:        tag,src_host
aggregate_filter[inbound]:  dst net 209.97.32.0/20 and src net not 209.97.32.0/20
aggregate_filter[outbound]: src net 209.97.32.0/20 and dst net not 209.97.32.0/20

interface:              eth0
interface_wait:         true
promisc:                true
syslog:                 local0
networks_file:          /app/pmCollector/etc/networks.lst
sql_host:               <snip>
sql_user:               <snip>
sql_passwd:             <snip>
sql_db:                 pmacct
sql_table:              acct_v2
sql_table_version:      2
sql_refresh_time:       300
sql_optimize_clauses:   true
sql_history:            5m
sql_history_roundoff:   m
sql_dont_try_update:    true
sql_recovery_logfile:   /app/pmacct-0.8.2/var/log/recover
post_tag:               1
sampling_rate:          1
--end--

Thanks,

Dave


On Wed, 2005-10-26 at 02:42 +0200, Paolo Lucente wrote:
Hello Dave,

On Tue, Oct 25, 2005 at 05:26:33PM -0400, David C. Maple wrote:

> We are using pmacctd to collect throughput in and out of our IPs.  The data is
> aggregated into 5 minute periods.  The first FULL 5 minute period is about 
> right: 5.6mb out, 400kb in.  The rest of the periods are up to 100 times the 
> first.  I've tried different configurations and included the relevant parts 
> of the latest here.  (I left out the database name and username/password stuff.)

I've compiled pmacct 0.9.3 with the --enable-mysql and --enable-64bit options on
a Fedora Core 3 and a RedHat 9. Both run a MySQL 4.0.14 standard. I've tested your
fragment of configuration (successfully) while backgrounding a download of a 650Mb
ISO from a RedHat server (with an almost constant transfer rate around 1.5Mbit/s):

mysql> select ip_src, ip_dst, stamp_inserted, stamp_updated, (bytes*8)/300 
	from acct_v4 where ip_src='150.145.84.133' or ip_dst='150.145.84.133';
----------------+----------------+---------------------+---------------------+---------------+
| ip_src         | ip_dst         | stamp_inserted      | stamp_updated       | (bytes*8)/300 |
+----------------+----------------+---------------------+---------------------+---------------+
|                | 150.145.84.133 | 2005-10-26 02:10:00 | 2005-10-26 02:15:01 |     324629.04 |
| 150.145.84.133 |                | 2005-10-26 02:10:00 | 2005-10-26 02:15:01 |       9868.64 |
|                | 150.145.84.133 | 2005-10-26 02:15:00 | 2005-10-26 02:20:01 |    1667962.88 |
| 150.145.84.133 |                | 2005-10-26 02:15:00 | 2005-10-26 02:20:01 |      33202.13 |
|                | 150.145.84.133 | 2005-10-26 02:20:00 | 2005-10-26 02:25:01 |    1483734.61 |
| 150.145.84.133 |                | 2005-10-26 02:20:00 | 2005-10-26 02:25:01 |      29866.13 |
|                | 150.145.84.133 | 2005-10-26 02:25:00 | 2005-10-26 02:30:01 |    1574592.00 |
| 150.145.84.133 |                | 2005-10-26 02:25:00 | 2005-10-26 02:30:01 |      31127.57 |
|                | 150.145.84.133 | 2005-10-26 02:30:00 | 2005-10-26 02:35:01 |    1889721.60 |
| 150.145.84.133 |                | 2005-10-26 02:30:00 | 2005-10-26 02:35:01 |      36796.69 |
|                | 150.145.84.133 | 2005-10-26 02:35:00 | 2005-10-26 02:40:01 |    1658758.93 |
| 150.145.84.133 |                | 2005-10-26 02:35:00 | 2005-10-26 02:40:01 |      32058.56 |
+----------------+----------------+---------------------+---------------------+---------------+

In the end, i'm able to compare the size reported by pmacct with the partial size
effectively downloaded on the disk. I'll leave it testing for the rest of the file
just to see how things get evolved (and will let you know).

Few questions for you: did you use the default (v2) table schema ? Are you sure you
have included all relevant parts of your configuration in the previous post (eg. any
'sql_preprocess' line) ? If you try to downgrade to 0.9.2 or 0.9.1, are you still able
to notice such odd behaviour ? Which version of MySQL server are you actually running ?
Is the server running a plain (SMP) x86 architecture ?  

Cheers,
Paolo


_______________________________________________
pmacct-discussion mailing list
[email protected]
http://muffin.area.ba.cnr.it/mailman/listinfo/pmacct-discussion

Reply via email to