Home / user-40958

benmmurphy's questions

Martin Hope
benmmurphy
Asked: 2010-06-18 09:46:58 +0800 CST
  • 1

From some networks requesting certain images on our webserver is very flakey. I've looked at tcpdumps on both sides and the server sends back part of the file and the client ACKs the TCP packet but the server never receives the ACK.

The servers view:

41  19.941136   212.169.34.114  209.20.73.85    TCP 52456 > http [SYN] Seq=0 Win=8192 Len=0 MSS=1460 WS=2
42  19.941136   209.20.73.85    212.169.34.114  TCP http > 52456 [SYN, ACK] Seq=0 Ack=1 Win=5440 Len=0 MSS=1360
46  20.041142   212.169.34.114  209.20.73.85    TCP 52456 > http [ACK] Seq=1 Ack=1 Win=65280 Len=0
47  20.045142   212.169.34.114  209.20.73.85    HTTP    GET /map/map/s+74-WBkWk0aR28Yy-YjXA== HTTP/1.1
48  20.045142   209.20.73.85    212.169.34.114  TCP http > 52456 [ACK] Seq=1 Ack=522 Win=6432 Len=0
49  20.045142   209.20.73.85    212.169.34.114  TCP [TCP segment of a reassembled PDU] (Part of the content of the image 2720 bytes. i assume it is reassembled in tcpdump and it is fragmented over the wire.)
** never receives the ACK sent in frame 282 and will eventually resend the tcp segment **

The clients view:

274 26.161773   10.0.16.67  209.20.73.85    TCP 52456 > http [SYN] Seq=0 Win=8192 Len=0 MSS=1460 WS=2
276 26.262867   209.20.73.85    10.0.16.67  TCP http > 52456 [SYN, ACK] Seq=0 Ack=1 Win=5440 Len=0 MSS=1360
277 26.263255   10.0.16.67  209.20.73.85    TCP 52456 > http [ACK] Seq=1 Ack=1 Win=65280 Len=0
278 26.265193   10.0.16.67  209.20.73.85    HTTP    GET /map/map/s+74-WBkWk0aR28Yy-YjXA== HTTP/1.1 
279 26.365562   209.20.73.85    10.0.16.67  TCP http > 52456 [ACK] Seq=1 Ack=522 Win=6432 Len=0
280 26.368002   209.20.73.85    10.0.16.67  TCP [TCP segment of a reassembled PDU] (Part of the content of the image. Only 1400 bytes.)
282 26.571380   10.0.16.67  209.20.73.85    TCP 52456 > http [ACK] Seq=522 Ack=1361 Win=65280 Len=0

The network we are having trouble with is NATd. Is there any kind of explanation for this weirdness?

nat blackhole
Martin Hope
benmmurphy
Asked: 2010-04-23 04:28:54 +0800 CST
  • 0

I'm using suidperl to run some programs that require root permissions. however, the runtime linker won't expand library paths which contain $ORIGIN entries so the programs i want to run (jstack from java) won't run.

more info here

There is one exception to the advice to make heavy use of $ORIGIN. The runtime linker will not expand tokens like $ORIGIN for secure (setuid) applications. This should not be a problem in the vast majority of cases.

my program looks something like this:

#!/usr/bin/perl
$ENV{PATH} = "/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin:/usr/java/jdk1.6.0_12/bin:/root/bin";
$ENV{JAVA_HOME} = "/usr/java/jdk1.6.0_12";
open(FILE, '/var/run/kil.pid');
$pid = <FILE>;
close(FILE);
chomp($pid);
if ($pid =~ /^(\d+)/) {
 $pid = $1;
} else {
  die 'nopid';
}

system( "/usr/java/jdk1.6.0_12/bin/jstack", "$pid");

is there any way to fork off a child process in a way so that the linker will work correctly.

linux
Martin Hope
benmmurphy
Asked: 2010-04-20 07:50:59 +0800 CST
  • 2

I have a setup where squid sits in front of a java server and acts as a reverse proxy. Recently i've load tested the site and if i fire 100 threads at it each making a request using jmeter i start getting errors in my load test tool like 'no route to host' even though the load test tool and the server are on the same machine.

if i run the following command where port 82 is the port my squid server is running on:

netstat -ann | grep 82 | wc -l

i get 22000 or something and most of them are in TIMED_WAIT. i'm thinking that maybe the huge number of sockets in the TIMED_WAIT state are starving the box of resources.

linux squid