Update: These scripts are no longer maintained. They have been superceded by the OpenWrtScripts repo
TL;DR These scripts test your network for bufferbloat, and configure your CeroWrt router.
-
Test your network for bufferbloat from any Linux or OSX computer with betterspeedtest.sh and netperfwrapper.sh Both scripts require the netperf benchmark program from http://netperf.org
-
Configure a CeroWrt Router to a repeatable state with config-cerowrt.sh, tunnelbroker.sh, and cerostats.sh.
The CeroWrt router firmware project has largely eliminated the problem of bufferbloat on Ethernet for home routers. This algorithm makes a huge difference for wireless, too, although there's still more work to be done.
Bufferbloat is undesired latency that happens when routers buffer too much data during periods of high traffic. The symptoms of bufferbloat makes "the Internet feel slow." The Smart Queue Management (SQM) techniques based on the fq_codel algorithm developed by the CeroWrt team are being widely adopted across the Internet to make everyone's network performance better.
We developed the scripts below (sometimes also called "Ceroscripts") to measure the improvements to latency in as we worked on CeroWrt. These scripts include:
-
Scripts that measure the performance of your router or offer load to the network for testing.
-
Script to configure the CeroWrt router consistently after flashing factory firmware.
-
Script to set up a IPv6 6-in-4 tunnel to TunnelBroker.net.
-
Script to collect troubleshooting information that helps us diagnose problems in the CeroWrt distribution.
These scripts are bundled into the final CeroWrt build as the 'cerowrtscripts' package, saved in the /usr/lib/CeroWrtScripts directory.
To get the newest versions, you can use opkg update; opkg upgrade
If the scripts are not built into your version of CeroWrt, it is safe to put them in that CeroWrtScripts directory.
This script emulates a web-based speedtest, but does it one better. During the download and upload, the script simultaneously measures latency of pings to see whether the data transfers affect the responsiveness of your network.
Here's why that's important: If the data transfers increase the latency/lag significantly, then other network activity, such as voice or video chat, gaming, and general network activity will also work poorly. Gamers will see this as lagging out when someone else uses the network. Skype and FaceTime will see dropouts or freezes. Latency is bad, and good routers will control it.
The betterspeedtest.sh script measures latency during data transfers. It will run on any computer where you can install the netperf software. To invoke it:
sh betterspeedtest.sh [ -4 | -6 ] [ -H netperf-server ] [ -t duration ] [ -p host-to-ping ] [-n simultaneous-streams ]
Options, if present, are:
- -H | --host: DNS or Address of a netperf server (default - netperf.bufferbloat.net)
Alternate servers are netperf-east (east coast US), netperf-west (California), and netperf-eu (Denmark) - -4 | -6: Enable ipv4 or ipv6 testing (default - ipv4)
- -t | --time: Duration for how long each direction's test should run - (default - 60 seconds)
- -p | --ping: Host to ping to measure latency (default - gstatic.com)
- -n | --number: Number of simultaneous sessions (default - 5 sessions)
The output shows separate (one-way) download and upload speed, along with a summary of latencies, including min, max, average, median, and 10th and 90th percentiles so you can get a sense of the distribution. The tool also displays the percent packet loss.
Example: The example below shows two measurements on a a 7mpbs/768kbps DSL circuit. On the left is a test run without SQM; on the right is a test using SQM.
-
Without SQM, upload latency gets huge (greater than 5 seconds), meaning that network performance would be terrible for everyone during that time. In this test, there doesn't seem to be much bloat in the download direction, although it happens in many other circumstances.
-
With SQM, transmit and receive rates remain comparable to the no-SQM case, while latency only goes up a little in either directions (less than 27 msec under load).
Test results:
Example with NO SQM - BAD Example using SQM - GOOD
root@cerowrt:/usr/lib/CeroWrtScripts# sh betterspeedtest.sh root@cerowrt:/usr/lib/CeroWrtScripts# sh betterspeedtest.sh
[date/time] Testing against netperf.bufferbloat.net (ipv4) [date/time] Testing against netperf.bufferbloat.net (ipv4)
with 5 simultaneous sessions while pinging gstatic.com with 5 simultaneous sessions while pinging gstatic.com
(60 seconds in each direction) (60 seconds in each direction)
Download: 6.84 Mbps Download: 6.91 Mbps
Latency: (in msec, 58 pings, 0.00% packet loss) Latency: (in msec, 61 pings, 0.00% packet loss)
Min: 37.982 Min: 38.696
10pct: 41.002 10pct: 42.565
Median: 57.647 Median: 54.135
Avg: 54.814 Avg: 54.097
90pct: 62.689 90pct: 62.777
Max: 64.680 Max: 65.401
Upload: 0.74 Mbps Upload: 0.76 Mbps
Latency: (in msec, 61 pings, 14.08% packet loss) Latency: (in msec, 53 pings, 0.00% packet loss)
Min: 39.213 Min: 38.241
10pct: 1404.113 10pct: 41.106
Median: 3392.531 Median: 46.484
Avg: 3244.456 Avg: 47.209
90pct: 4546.094 90pct: 52.409
Max: 5055.858 Max: 58.272
This script runs several netperf commands simultaneously. This mimics the stress test of Flent (formerly known as netperf-wrapper) but without the nice GUI result.
When you start this script, it concurrently uploads and downloads several streams (files) to a server on the Internet. This places a heavy load on the bottleneck link of your network (probably your connection to the Internet), and lets you measure both the total bandwidth and the latency of the link during the transfers.
To invoke the script:
sh netperfrunner.sh [ -4 | -6 ] [ -H netperf-server ] [ -t duration ] [ -p host-to-ping ] [-n simultaneous-streams ]
Options, if present, are:
- -H | --host: DNS or Address of a netperf server (default - netperf.bufferbloat.net)
Alternate servers are netperf-east (east coast US), netperf-west (California), and netperf-eu (Denmark) - -4 | -6: Enable ipv4 or ipv6 testing (default - ipv4)
- -t | --time: Duration for how long each direction's test should run - (default - 60 seconds)
- -p | --ping: Host to ping to measure latency (default - gstatic.com)
- -n | --number: Number of simultaneous sessions (default - 4 sessions)
The output of the script looks like this:
root@cerowrt:/usr/lib/CeroWrtScripts# sh netperfrunner.sh
[date/time] Testing netperf.bufferbloat.net (ipv4) with 4 streams down and up
while pinging gstatic.com. Takes about 60 seconds.
Download: 5.29 Mbps
Upload: 0.39 Mbps
Latency: (in msec, 64 pings, 1.54% packet loss)
Min: 37.493
10pct: 41.625
Median: 61.724
Avg: 63.557
90pct: 84.536
Max: 118.194
Note: The download and upload speeds reported may be considerably lower than your line's rated speed. This is not a bug, nor is it a problem with your internet connection. That's because the acknowledge messages for one stream can consume a significant fraction (as much as 25%) of the link's capacity in the opposite direction.
This script continually invokes the netperfrunner script to provide a long-running heavy load. It runs forever - Ctl-C will interrupt it.
This script updates the factory settings of CeroWrt to a known-good configuration. If you frequently update your firmware, you can use this script to reconfigure the router to a consistent state. You should make a copy of this script, customize it to your needs, then use the "To run this script" procedure (below).
This script is designed to configure the settings after an initial "factory" firmware flash. There are sections below to configure many aspects of your router. All the sections are commented out. There are sections for:
- Set up the ge00/WAN interface to connect to your provider
- Update the software packages
- Update the root password
- Set the time zone
- Enable SNMP for traffic monitoring and measurements
- Enable NetFlow export for traffic analysis
- Enable mDNS/ZeroConf on the ge00 (WAN) interface
- Change default IP addresses and subnets for interfaces
- Change default DNS names
- Set the SQM (Smart Queue Management) parameters
- Set the radio channels
- Set wireless SSID names
- Set the wireless security credentials
To run this script
Flash the router with factory firmware. Then ssh in and execute these statements. You should do this over a wired connection because some of these changes may reset the wireless network.
ssh root@172.30.42.1
cd /tmp
cat > config.sh
[paste in the contents of this file, then hit ^D]
sh config.sh
Presto! (You should reboot the router when this completes.)
Note: If you use a secondary CeroWrt router, you can create another copy of this script, and use it to set different configuration parameters (perhaps different subnets, radio channels, SSIDs, enable mDNS, etc).
This script configures CeroWrt to create an IPv6 tunnel. It's an easy way to become familiar with IPv6 if your ISP doesn't offer native IPv6 capabilities. There are three steps:
-
Go to the Hurricane Electric TunnelBroker.net site to set up your free account. There are detailed instructions for setting up an account and an IPv6 tunnel at the CeroWrt IPv6 Tunnel page.
-
Edit the tunnelbroker.sh script, using the parameters supplied by Tunnelbroker.net. They're on the site's "Tunnel Details" page. Click on the "Example Configurations" tab and select "OpenWRT Backfire 10.03.1". Use the info to fill in the corresponding lines of the script.
-
ssh into the CeroWrt router and execute this script with these steps.
ssh root@172.30.42.1 cd /tmp cat > tunnel.sh [paste in the contents of this file, then hit ^D] sh tunnel.sh [Restart your router. This seems to make a difference.]
Presto! Your tunnel is up! Your computer should get a global IPv6 address, and should be able to communicate directly with IPv6 devices on the Internet. To test it, try: ping6 ivp6.google.com
This script collects a number of useful configuration settings and dynamic values for aid in diagnosing problems with CeroWrt. If you report a problem, it would be helpful to include the output of this script.
By default, it collects information about the first 2.4GHz radio/interface, and writes the collected data to /tmp/cerostats_output.txt