Re: [Proftpd-user] Problem with a lot of connections

Hi,

On Mon, Nov 30, Göran Hasse wrote:

> I have a server, Ubuntu, (5.4.0-53-generic) with ProFTP (ProFTPD 1.3.6c).
> 
> On this we have a LOT of IoT devices logging in at the SAME account. There is
> about > 100 of those devices connecting at exactly(!) the same time. They run
> on a GPS clock and the connections is within ms. 
> 
> Very often they can connect, login, create the file to be transfered,
> and then deliver data but sometimes they can't deliver data.
> 
> One potential problem is that our net operator is NAT:ing all our devices
> and they get the same IP-nummber. And this could be a resource problem
> in their NAT software. One other potential problem is ProFTPD cand handle
> this many passive FTP connections.
> 
> <pre>
> 2020-11-28 07:10:15,890 ajabsab proftpd[890999] ajabsab (193.181.246.218[193.181.246.218]): Data transfer stall timeout: 600 seconds
> 2020-11-28 07:10:15,891 ajabsab proftpd[890999] ajabsab (193.181.246.218[193.181.246.218]): Transfer aborted after 0 bytes in 583.64 seconds
> 2020-11-28 07:10:15,891 ajabsab proftpd[890999] ajabsab (193.181.246.218[193.181.246.218]): notice: user ftp-incomming: aborting transfer: Data connection closed
> 2020-11-28 07:10:15,891 ajabsab proftpd[890999] ajabsab (193.181.246.218[193.181.246.218]): dispatching POST_CMD_ERR command 'STOR data.json' to mod_exec
> 2020-11-28 07:10:15,891 ajabsab proftpd[890999] ajabsab (193.181.246.218[193.181.246.218]): dispatching POST_CMD_ERR command 'STOR data.json' to mod_quotatab
> 2020-11-28 07:10:15,891 ajabsab proftpd[890999] ajabsab (193.181.246.218[193.181.246.218]): dispatching LOG_CMD_ERR command 'STOR data.json' to mod_log
> 2020-11-28 07:10:15,891 ajabsab proftpd[890999] ajabsab (193.181.246.218[193.181.246.218]): dispatching LOG_CMD_ERR command 'STOR data.json' to mod_xfer
> 2020-11-28 07:10:15,891 ajabsab proftpd[890999] ajabsab (193.181.246.218[193.181.246.218]): FTP session closed.
> 2020-11-28 07:10:20,904 ajabsab proftpd[890755] ajabsab (193.181.246.218[193.181.246.218]): FTP no transfer timeout, disconnected
> 2020-11-28 07:10:20,905 ajabsab proftpd[890755] ajabsab (193.181.246.218[193.181.246.218]): FTP session closed.
> 2020-11-28 07:10:21,489 ajabsab proftpd[890795] ajabsab (193.181.246.218[193.181.246.218]): FTP no transfer timeout, disconnected
> 2020-11-28 07:10:21,490 ajabsab proftpd[890795] ajabsab (193.181.246.218[193.181.246.218]): FTP session closed.
> 2020-11-28 07:12:23,997 ajabsab proftpd[890817] ajabsab (193.181.246.218[193.181.246.218]): FTP no transfer timeout, disconnected
> 2020-11-28 07:12:23,998 ajabsab proftpd[890836] ajabsab (193.181.246.218[193.181.246.218]): FTP no transfer timeout, disconnected
> 2020-11-28 07:12:23,998 ajabsab proftpd[890817] ajabsab (193.181.246.218[193.181.246.218]): FTP session closed.
> 2020-11-28 07:12:23,999 ajabsab proftpd[890836] ajabsab (193.181.246.218[193.181.246.218]): FTP session closed.
> 2020-11-28 07:16:17,789 ajabsab proftpd[891183] ajabsab (193.181.246.218[193.181.246.218]): FTP no transfer timeout, disconnected
> 2020-11-28 07:16:17,790 ajabsab proftpd[891183] ajabsab (193.181.246.218[193.181.246.218]): FTP session closed.
> </pre>

maybe you can insert a random sleep before the ftp command is executed.

in Perl I did it this way:

--snip--
use Time::HiRes  qw( usleep );
my $maxtimewait  = 10000000;  # random micro seconds we will wait before sending data

....

# we will wait some random time, so not all servers will send data at the very same time
my $waittime = int ( rand ( $maxtimewait ) );
print_debug ( "will wait for $waittime microseconds" );
usleep ( $waittime );

I think you can insert random waits in every language ....


-- 
Gruß

  Dieter

--
I do not get viruses because I do not use MS software.
If you use Outlook then please do not put my email address in your
address-book so that WHEN you get a virus it won't use my address in the
From field.


_______________________________________________
ProFTPD Users List   <[email protected]>
Unsubscribe problems?
http://www.proftpd.org/list-unsub.html