I'll do my best to get both ends, the client is on read-only flash, so it makes that end difficult.

But, as a general question, it has no problem handling the load if it is straight udp. So, I imagine this is an issue with either encryption, or, most likely, TCP. Do you have good pointers on the various syslog tuning params for TCP?

thx

On 1/23/10 1:17 AM, Zoltán Pallagi wrote:
Ok, then show the debug logs of server syslog-ng when you restart rsyslog, please. You may need to check the debug logs of rsyslog as well (I've never used rsyslog, but you can more information on the rsyslog webpage: http://www.rsyslog.com/doc-troubleshoot.html)

I would like to help you, but I can't find out the reason for this behavior without any information, so please tell me as much information as possible.

2010.01.23. 0:02 keltezéssel, Rory Toma írta:
I am not using syslog-ng on the client. I am using rsyslog.

On 1/22/10 2:57 PM, Zoltán Pallagi wrote:
Syslog-n tries to reconnect in every 60th seconds to server by default. Perhaps, the first attempt was not succes and you need wait for the second attempt so it can take 1-2 minutes.
If you reduce the "time_reopen()" on your client's configuration, it can be faster.

I also have an other theory, if you are using flow-control on client side, and the server is a bit overloaded, syslog-ng can stop reading the source (and sending to the server).

Just show a few debug lines after starting syslog-ng on client and a few lines on server in the same time. (because syslog-ng will send a few log about the state of connections, e.g.: accepted or closed)


2010.01.22. 22:11 keltezéssel, Rory Toma írta:
The problem has been alleviated somewhat by moving to a 64-bit platform. 
Since these generally have a different set of tcp defaults, there is 
probably a tcp value(s) that need tuning. I do notice on this one, that 
after I reboot a client, it takes a few minutes for the logs to start 
flowing, but flow they do.

I can show you the output, but it's quite long.

On 1/22/10 1:26 AM, Zoltán Pallagi wrote:
  
Hi,

Can you show me the output of "syslog-ng -Fevd" on your client after 
restarting?

(you can find more information about debugging syslog-ng on my blog: 
http://pzolee.blogs.balabit.com/)

Rory Toma írta:
    
I am using syslog-ng-3.0.4-1.rhel on a CentOS-5.4 system. I am using 
tls, and have a setup as below. Here is what happens. It logs fine. 
However, if I reboot my client that is sending logs, it no longer 
works until I restart the syslog-ng server. What do I need to do here?

@version: 3.0
options { flush_lines (3);
           time_reopen (10);
           log_fifo_size (1000);
           long_hostnames (off);
           use_dns (no);
           use_fqdn (no);
           create_dirs (yes); dir_perm (0755);
           keep_hostname (yes);
           ts_format("iso");
         };

source telo {
         tcp( port(80)
         tls( key_file("/export/tls/key.pem")
              cert_file("/export/tls/cert.pem")
              peer_verify(optional-untrusted)) ); };

# Myx destinations
destination myx_dest_0000 { 
file("/logs/myx_008161000/$R_YEAR$R_MONTH$R_DAY/$HOS
T-$R_YEAR$R_MONTH$R_DAY.log" owner(root) group(root) perm(0644) 
template("$YEAR-
$MONTH-$DAY $HOUR:$MIN:$SEC $MSG\n") template_escape(no)); };
filter myx_filter_0000 { host("myx_001861000[0-9A-F]\{3\}$"); };
log { source(telo); filter(myx_filter_0000); 
destination(myx_dest_0000); };



______________________________________________________________________________ 

Member info: https://lists.balabit.hu/mailman/listinfo/syslog-ng
Documentation: 
http://www.balabit.com/support/documentation/?product=syslog-ng
FAQ: http://www.campin.net/syslog-ng/faq.html


      
    
______________________________________________________________________________
Member info: https://lists.balabit.hu/mailman/listinfo/syslog-ng
Documentation: http://www.balabit.com/support/documentation/?product=syslog-ng
FAQ: http://www.campin.net/syslog-ng/faq.html


  


--
pzolee



--
pzolee