[tcpdump-workers] Latency impact while listening at eth0, lo and any on Linux
Hello, I use pcap library to develop a tool that measures on the fly network latency (a time between receiving a request and sending a response) for HTTP, DIAMETER and other protocols. My tool is compiled for x86_64 Linux , So I noticed significant latency growing if my tool listening traffic at 'any' interface in compare with listening at 'eth0'. Listening at 'lo' also increases latency. So I wander why does latency increase and can I do something to reduce latency impact while capturing traffic at 'any' or 'lo' interfaces? Regards, Sergey Gomanyuk This message and the information contained herein is proprietary and confidential and subject to the Amdocs policy statement, you may review at http://www.amdocs.com/email_disclaimer.asp ___ tcpdump-workers mailing list tcpdump-workers@lists.tcpdump.org https://lists.sandelman.ca/mailman/listinfo/tcpdump-workers
[tcpdump-workers] Can libpcap affect speed of packets sending/recieving?
Hi, I wonder if libpcap can somehow affect speed of processing packets by TCP/IP stack? I noticed that such behavior can happen in particular Linux environments with high traffic load - without capturing all is Ok, with capturing (especially on "any" interface) it seems that packets stuck in TCP/IP stack (sometimes application that generates traffic receives a huge amount of EAGAIN error). Thanks, Sergey Gomanyuk This message and the information contained herein is proprietary and confidential and subject to the Amdocs policy statement, you may review at http://www.amdocs.com/email_disclaimer.asp ___ tcpdump-workers mailing list tcpdump-workers@lists.tcpdump.org https://lists.sandelman.ca/mailman/listinfo/tcpdump-workers
