[syslog-ng]syslog-ng doesn't re-open the TCP connection
Rosenberg, Relu (Exchange)
rrosenberg@bear.com
Tue, 7 May 2002 18:47:01 -0400
This is syslog-ng 1.4.13 on Solaris 8, reading from named pipe and sending
over tcp to a central server syslog-ng.
Configuration file:
options {
long_hostnames (on) ;
use_dns (no) ;
use_fqdn (yes) ;
keep_hostname (yes) ;
chain_hostnames (no) ;
log_fifo_size(1000) ;
} ;
source s_pipe {
pipe ("/tmp/.syslog-ng.pipe") ;
internal () ;
} ;
destination d_tcp {
tcp (remote_host port(2222)) ;
} ;
log {
source (s_pipe) ;
destination (d_tcp) ;
} ;
In order not to replace the syslog I tail the /var/adm/messages and redirect
into the /tmp/.syslog-ng.pipe.
Syslog-ng is reading the pipe and sending to remote syslog-ng over tcp.
At a point in time it stopped sending messages, doesn't try to re-open the
lost TCP connection. The connection may have been lost due to network
outage or maintenance work, but it should reopen.
This happens to over 10 clients, I have over 300 working this way.
The truss output shows the read of new messages from named pipe, but no
attempt to write, or to re-open the connection to the remote server
51/1: poll(0xFFBEFB10, 2, 60000) = 1
351/1: read(3, " M a y 7 1 8 : 2 8".., 1024) = 70
351/1: time() = 1020810505
351/1: time() = 1020810505
351/1: time() = 1020810505
351/1: time() = 1020810505
351/1: poll(0xFFBEFB10, 2, 100) = 0
351/1: poll(0xFFBEFB10, 2, 60000) (sleeping...)
351/2: signotifywait() (sleeping...)
351/3: lwp_cond_wait(0xFF0F55B0, 0xFF0F55C0, 0xFF1F5C48)
(sleeping...)
thanks
Relu Rosenberg
****************************************************************
Bear Stearns is not responsible for any recommendation, solicitation,
offer or agreement or any information about any transaction, customer
account or account activity contained in this communication.
***********************************************************************