Splunk Community,
I have a Netscaler appliance configured to send syslog data to a syslog-ng server over TCP/9524. The TCP connection is reaching the sylog-ng server but never completes the handshake.
Netscaler --> Syslog-ng (SYN)
Syslog-ng --> Netscaler (SYN, ACK)
Netscaler --> Syslog-ng (FIN, ACK)
Syslog-ng --> Netscaler (FIN, ACK)
Netscaler --> Syslog-ng (ACK)
We have multiple TCP connections on various other ports connecting fine to this Syslog-ng server. Wondering if anyone else has had this experience or suggestions to remediate. Most appreciated.
Thank you,
Tom
Hi,
Try running tcpdump on the server to see if any netscaler messages reach the server (but fail to reach syslog-ng for some reason), and try running syslog-ng in verbose mode to see if it shows any error messages.
Thanks Frobert. This is from a tcpdump on the syslog-ng server,
Netscaler --> Syslog-ng (SYN)
Syslog-ng --> Netscaler (SYN, ACK)
Netscaler --> Syslog-ng (FIN, ACK)
Syslog-ng --> Netscaler (FIN, ACK)
Netscaler --> Syslog-ng (ACK)
The three-way handshake never completes. I am researching as a possible source of the issue a mismatch in the MSS of the TCP Packet.
In the SYN packet, an MSS of 1380 is sent,
TCP Option - Maximum segment size: 1380 bytes
The syslog-ng server responds with an MSS of 1460
TCP Option - Maximum segment size: 1460 bytes
The very next packet is the FIN,ACK. My theory right now is that the two servers cannot agree on the correct MSS and am researching with the Netscaler admin.
Thank you,