tcpcopy – an online request replication tool, fit for online testing, stress testing, performance evaluation,etc

It can help you find bugs in your online project without actually being online. And it can also be used to test the stress that a system could endure.For example, if your system already has “memcached” subsystem and you want to  use “membase” to replace it, tcpcopy can assist you to test “membase“. While your old memcached system is still running online, tcpcopy could copy the flow of packets from memcached to membase. From the point view of membase, the flow is accessing membase(just like membase online), and it will not affect memcached at all except  network bandwidth and a little cpu load.

1) Distributed Stress Test
You can use online data to test the stress that your target machine can endure. It is better
than apache ab tool and you can find bugs that only occur during high-stress situations.
2) Hot Backup
It is very suitable for backup tasks if connections are short-lived and the request loss rate
is very low(1/100000).
3) Normal Online Test
You can find whether the new system is stable and find bugs that only occur in actual
online environments.
4) Comparison Test
For example, you can use tcpcopy to compare the performances of apache and nginx.

1)real time
4)easy to use

How to use tcpcopy?
preparing steps:
1) download tar.gz file
4)make install

running steps:
the target machine(root privilege):
modprobe ip_queue (if not running)
iptables -I OUTPUT -p tcp –sport port -j QUEUE (if not set)
online machine(root privilege):
./tcpcopy local_ip1[:local_ip2:…]  local_port  remote_ip  remote_port

An example:
Suppose 13 and 14 are online machines, 148 is a target machine which is similar to
the online machines, and 12321 is used both as local port and remote port. We
use tcpcopy to test if 148 can endure two times of current online stress.

Using tcpcopy to perform the above test task.
the target machine(148)
# modprobe ip_queue (if not run up)
# iptables -I OUTPUT -p tcp –sport 12321 -j QUEUE (if not set)
# ./interception
online machine(13):
# ./tcpcopy xx.xx.xx.13 12321 xx.xx.xx.148 12321
online machine(14):
# ./tcpcopy xx.xx.xx.14 12321 xx.xx.xx.148 12321

Cpu load and memory usage are as follows:
13 cpu:
11124 adrun 15 0 193m 146m 744 S 18.6 7.3 495:31.56 asyn_server
11281 root 15 0 65144 40m 1076 S 12.3 2.0 0:47.89 tcpcopy
14 cpu:
16855 adrun 15 0 98.7m 55m 744 S 21.6 2.7 487:49.51 asyn_server
16429 root 15 0 41156 17m 1076 S 14.0 0.9 0:33.63 tcpcopy
148 cpu :
25609 root 15 0 76892 59m 764 S 49.6 2.9 63:03.14 asyn_server
20184 root 15 0 5624 4232 292 S 17.0 0.2 0:52.82 interception

Access log analysis:
13 online machine: grep ‘Tue 11:08’ access_0913_11.log |wc -l :89316,  1489 reqs/sec
14 online machine: grep ‘Tue 11:08’ access_0913_11.log |wc -l :89309,  1488 reqs/sec
148 test  machine: grep ‘Tue 11:08’ access_0913_11.log |wc -l :178175, 2969 reqs/sec
request loss rate:

From the above, we can see that the target machine can endure two times of current
online stress.

  What about the cpu load ?
tcpcopy on online machine 13 occupies 12.3% of cpu load, tcpcopy on online 14
occupies 14%,and interception on target machine 148 occupies 17%. We can see
that the cpu load is very low here, and so is the memory usage.

1) It is tested on machines running linux (kernal 2.6 or above).
2) tcpcopy is similar to udp, so sometimes it may lose requests.
3) If you tcpcopy local request, please set lo MTU not more than 1500.
4) Interception(tcpcopy server) is single threaded.
5) tcpcopy needs root privilege.
6) document
7) Check error.log if you have trouble and email to (
8) long request(such as uploading a big file) is not working well here(no retransmission
when sending packets to the target test server)

Download Latest Version : tcpcopy-0.4.0.tar.gz (394 KB)
Find Other version | 
Read more For references : |