Pktgen for CI

ci/.gitignore

@@ -1,4 +1,3 @@
-access_log
 config
 webhook-config.json
 encrypted_secret.bin
@@ -10,4 +9,7 @@ repository
 linter-output.txt
 *key
 *.pub
-*.stats
+*stats
+*out*
+*log*
+nimbnode*

ci/README.md

@@ -69,7 +69,7 @@ The CI process can be broken into multiple steps:
 
 5. Run linter on the checked out code
 
-    Runs the `run_linter` function in `helper-functions.sh`
+    Runs the `run_linter` function in `helper-manager-functions.sh`
 
 6. Clean up and restart all worker nodes
 
@@ -83,11 +83,15 @@ The CI process can be broken into multiple steps:
 
     Use paramiko to ssh and run `run-workload.py`
 
-9. Acquire results from the worker nodes
+9. Run modes are supplied to tell the worker which applications to test
+
+     Handle installation with `worker_files/worker.sh` for builds, and setting up manager for performance tests
+
+10. Acquire results from the worker nodes
 
     Use scp to copy the result stat file from worker
 
-10. Submit results as a comment on github
+11. Submit results as a comment on github
 
     Uses the `post-msg.py` script
 
@@ -106,6 +110,33 @@ ProxyPassReverse /onvm-ci/ http://nimbnode44:8080/
 ```  
 (Also need to setup github webhook to post to **http://nimbus.seas.gwu.edu/onvm-ci/github-webhook**)
 
+### Public and Private CI Runs
+
+CI is now able to accept requests from unauthenticated users. There is a list of Github users in the public project allowed to create a full run. Anyone who is able to view the private `-dev` repository is able to run CI there as well. In `openNetVM`, if a user is not in our list, the linter and branch checks will be executed, ignoring statistics calculations from the worker nodes.
+
+### Setting Up a Connected Worker
+
+Connecting two nodes is useful for measuring statistics with tools like Pktgen and the MTCP stack. There is a bit of setup required to get working connection working. Firstly, an SFP+ 10Gb Intel cable will be required to connect the Network Interface Cards in the two machines. Once this is done, attempt to bring up the correct interfaces for a stable connection. Some debugging might be required:  
+- If you don't know which `ifconfig -a` interface is correct, use `ethtool -p <interface name> 120`  
+  - This will blink a light on the interface (you have to be next to the machine for this to help)
+- Do this on both machines, to find the name of the interfaces that are linked
+- Run `sudo ifconfig <interface name> 11.0.0.1/24 up` on the first machine and `sudo ifconfig <interface name> 11.0.0.2/24 up`
+  - This will ensure `ping` understands what IP address it is supposed to talk to
+- If `ping -I <interface> 11.0.0.2` on the first machine works, great, if not, try changing the IP addresses or viewing `dmesg`
+
+Now that the interfaces are connected, choose which machine will be the CI worker, and which is a helper (Pktgen for example). Install Pktgen on this node by sending the `ci/install_pktgen` files to that machines' home folder. *Remember public keys must be created for all new machines*. Store these public keys in a folder with the server name, see the next section on statistics for more information. Run `chmod +x install-pktgen.sh` if it's not already an executable and run `./install-pktgen.sh` to install everything. If there are dependency errors, the machine might be a different version, so try to install the necessary packages. Once everything is installed, test ONVM->Pktgen between the machines, and if a connection is established, CI should work just fine with no more setup!
+
+### Advanced Statistics
+
+As CI continued to improve, with more programs to test with, benchmarks were made to track the average performance of a worker. In the future, CI will be able to handle multiple workers running many different tests. Since server configurations are not all the same, some with different hardware (Intel x710 vs. x520 NIC for example), performance of the nodes will not be the same. All that matters with CI is that the result of a run is the same or better, not globally across all nodes, but based on the specific server it ran on. For each worker, create a folder in the ci directory with the name of the worker IP. For example if `nimbnode17` is the current worker, a folder with path `/ci/nimbnode17/` should exist. In this folder, 3 files should be there at least. Firstly, a `benchmarks` file (used by the manager) should look similar to this: 
+
+```
+AVG_SPEED_TESTER_SPEED=40000000
+AVG_PKTGEN_SPEED=10000000
+AVG_MTCP_SPEED=.230
+```
+This is a configuration file, sourced by the manager to keep track of `nimbnode17`'s average performance for each test (currently Speed Tester, Pktgen, and mTCP). The other two files in the folder should be the two public keys, one for the worker, and the second for the worker's client server. Check the previous section on setting up a connection for more information.
+
 ### Checking if Online
 
 If you are worried if the CI build is offline or want to make sure it is listening for events, you can check the following url: curl http://nimbus.seas.gwu.edu/onvm-ci/status.  If that URL returns 404, CI is offline.  Otherwise it will display a message saying it is online.

ci/ci_busy.sh

@@ -2,7 +2,7 @@
 
 set -e
 
-. helper-functions.sh
+. helper-manager-functions.sh
 SCRIPT_LOC=$(pwd)
 
 print_header "Validating Config File and Sourcing Variables"

ci/helper-functions.sh → ci/helper-manager-functions.sh

@@ -38,45 +38,6 @@ print_header() {
     echo ""
 }
 
- # sets up dpdk, sets env variables, and runs the install script
-install_env() {
-    git submodule sync 
-    git submodule update --init
-
-    echo export ONVM_HOME=$(pwd) >> ~/.bashrc
-    export ONVM_HOME=$(pwd)
-
-    cd dpdk
-
-    echo export RTE_SDK=$(pwd) >> ~/.bashrc
-    export RTE_SDK=$(pwd)
-
-    echo export RTE_TARGET=x86_64-native-linuxapp-gcc  >> ~/.bashrc
-    export RTE_TARGET=x86_64-native-linuxapp-gcc
-
-    echo export ONVM_NUM_HUGEPAGES=1024 >> ~/.bashrc
-    export ONVM_NUM_HUGEPAGES=1024
-
-    echo $RTE_SDK
-
-    sudo sh -c "echo 0 > /proc/sys/kernel/randomize_va_space"
-
-    cd ../
-    pwd
-    . ./scripts/install.sh
-}
-
-# makes all onvm code
-build_onvm() {
-    cd onvm
-    make clean && make
-    cd ../
-
-    cd examples
-    make clean && make
-    cd ../
-}
-
 # obtains core config in cores.out file
 obtain_core_config() {
     cd scripts
@@ -140,3 +101,9 @@ run_linter() {
         fi
     done
 }
+
+# inputs are key_file, worker ip address, stats file - in that order 1,2,3
+fetch_files() {
+    scp -i $1 -oStrictHostKeyChecking=no -oUserKnownHostsFile=/dev/null $2:$3 ./$2.$3
+    check_exit_code "ERROR: Failed to fetch results from $2"
+}

ci/install_pktgen/helper-worker-functions.sh

@@ -0,0 +1 @@
+../worker_files/helper-worker-functions.sh

ci/install_pktgen/install-pktgen.sh

@@ -0,0 +1,41 @@
+#!/bin/bash
+
+. helper-install-functions.sh
+
+set -e
+
+sudo rm -rf repository
+
+git clone https://github.com/sdnfv/openNetVM.git repository
+check_exit_code "ERROR: Failed cloning"
+
+print_header "Installing Dependencies"
+sudo apt-get update
+sudo apt-get upgrade -y
+sudo apt-get install -y build-essential linux-headers-$(uname -r) git 
+sudo apt-get install -y libnuma1
+sudo apt-get install -y libnuma-dev
+sudo apt-get install libpcap-dev
+sudo apt-get install libreadline-dev
+
+print_header "Installing Lua"
+curl -R -O http://www.lua.org/ftp/lua-5.3.5.tar.gz
+tar zxf lua-5.3.5.tar.gz
+cd lua-5.3.5
+sudo make linux test
+sudo make install
+
+cd repository
+
+print_header "Installing Environment"
+install_env $RUN_PKT
+check_exit_code "ERROR: Installing environment failed"
+
+print_header "Make pktgen-dpdk"
+cd ~/repository/tools/Pktgen/pktgen-dpdk/
+make
+
+print_header "Updating lua script"
+cp ~/pktgen-timed-config.lua ~/repository/tools/Pktgen/openNetVM-Scripts/pktgen-config.lua
+
+print_header "Pktgen installed"

ci/install_pktgen/pktgen-timed-config.lua

@@ -0,0 +1,110 @@
+--                    openNetVM
+--      https://github.com/sdnfv/openNetVM
+--
+-- BSD LICENSE
+--
+-- Copyright(c)
+--          2015-2016 George Washington University
+--          2015-2016 University of California Riverside
+-- All rights reserved.
+
+-- Redistribution and use in source and binary forms, with or without
+-- modification, are permitted provided that the following conditions
+-- are met:
+
+-- Redistributions of source code must retain the above copyright
+-- notice, this list of conditions and the following disclaimer.
+-- Redistributions in binary form must reproduce the above copyright
+-- notice, this list of conditions and the following disclaimer in
+-- the documentation and/or other materials provided with the
+-- distribution.
+-- The name of the author may not be used to endorse or promote
+-- products derived from this software without specific prior
+-- written permission.
+
+-- THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+-- "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+-- LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+-- A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+-- OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+-- SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+-- LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+-- DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+-- THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+-- (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+-- OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+-- Change any of the settings below to configure Pktgen-DPDK
+
+-- A list of the test script for Pktgen and Lua.
+-- Each command somewhat mirrors the pktgen command line versions.
+-- A couple of the arguments have be changed to be more like the others.
+
+package.path = package.path ..";?.lua;test/?.lua;app/?.lua;"
+
+require "Pktgen"
+
+local function doWait(port, waitTime)
+    local idx;
+
+   pktgen.delay(1000);
+
+   pkt_rate_file = io.open("port_stats", "w");
+
+   if ( waitTime == 0 ) then
+       return;
+   end
+   waitTime = waitTime - 1;
+
+    -- Try to wait for the total number of packets to be sent.
+    local idx = 0;
+    while( idx < waitTime ) do
+	    -- Write port stats to output file separated by line
+        pkt_rate_file:write(pktgen.portStats("all", "rate")[0]["pkts_rx"] .. "\n");
+        idx = idx + 1;
+
+        local sending = pktgen.isSending(port);
+        if ( sending[tonumber(port)] == "n" ) then
+            break;
+        end
+        pktgen.delay(1000);
+    end
+
+    pkt_rate_file:close()
+end
+
+printf("Lua Version      : %s\n", pktgen.info.Lua_Version);
+printf("Pktgen Version   : %s\n", pktgen.info.Pktgen_Version);
+printf("Pktgen Copyright : %s\n", pktgen.info.Pktgen_Copyright);
+
+prints("pktgen.info", pktgen.info);
+
+printf("Port Count %d\n", pktgen.portCount());
+printf("Total port Count %d\n", pktgen.totalPorts());
+
+
+-- set up a mac address to set flow to 
+--  
+-- TO DO LIST:
+--
+-- Please update this part with the destination mac address, source and destination ip address you would like to sent packets to 
+
+pktgen.set_mac("0", "90:e2:ba:5e:73:21"); 
+pktgen.set_ipaddr("0", "dst", "10.11.1.17");
+pktgen.set_ipaddr("0", "src", "10.11.1.16");
+
+pktgen.set_proto("all", "udp");
+pktgen.set_type("all", "ipv4");
+
+pktgen.set("all", "size", 64)
+pktgen.set("all", "burst", 32);
+pktgen.set("all", "sport", 1234);
+pktgen.set("all", "dport", 1234);
+pktgen.set("all", "count", 1000000000);
+pktgen.set("all", "rate",100);
+
+pktgen.vlan_id("all", "start", 1);
+
+pktgen.start("all");
+doWait("all", 30);
+pktgen.quit();

ci/manager.sh

@@ -4,7 +4,7 @@
 set -e
 
 # source helper functions file
-. helper-functions.sh
+. helper-manager-functions.sh
 SCRIPT_LOC=$(pwd)
 
 print_header "Validating Config File and Sourcing Variables"
@@ -78,7 +78,7 @@ fi
 print_header "Cleaning up Old Results"
 
 sudo rm -f *.txt
-sudo rm -rf stats
+sudo rm -rf *stats
 sudo rm -rf repository
 
 print_header "Checking Worker and GitHub Creds Exist"
@@ -127,7 +127,6 @@ then
 fi
 
 print_header "Preparing Workers"
-
 for worker_tuple in "${WORKER_LIST[@]}"
 do
     tuple_arr=($worker_tuple)
@@ -157,10 +156,16 @@ do
     tuple_arr=($worker_tuple)
     worker_ip="${tuple_arr[0]}"
     worker_key_file="${tuple_arr[1]}"
-    scp -i $worker_key_file -oStrictHostKeyChecking=no -oUserKnownHostsFile=/dev/null -r ./repository $worker_ip:
-    check_exit_code "ERROR: Failed to copy ONVM files to $worker_ip"
-    scp -i $worker_key_file -oStrictHostKeyChecking=no -oUserKnownHostsFile=/dev/null helper-functions.sh worker.sh $worker_ip:
+    # make sure the config file is updated with the correct run mode
+    sed -i "/WORKER_MODE*/c\\WORKER_MODE=\"${RUN_MODE}\"" worker_files/worker-config
+    # create directory for scp
+    mkdir temp
+    # put all files in one temporary folder for one scp
+    cp -r ./$worker_ip/* ./repository ./worker_files/* temp
+    scp -i $worker_key_file -oStrictHostKeyChecking=no -oUserKnownHostsFile=/dev/null -r ./temp/* $worker_ip: 
     check_exit_code "ERROR: Failed to copy ONVM files to $worker_ip"
+    # get rid of the temp folder now for next worker
+    sudo rm -rf temp
 done
 
 print_header "Running Workloads on Workers"
@@ -174,18 +179,30 @@ do
 done
 
 print_header "Obtaining Performance Results from all workers"
-
-rm -f results_summary.stats
-
 for worker_tuple in "${WORKER_LIST[@]}"
 do
     tuple_arr=($worker_tuple)
     worker_ip="${tuple_arr[0]}"
     worker_key_file="${tuple_arr[1]}"
-    scp -i $worker_key_file -oStrictHostKeyChecking=no -oUserKnownHostsFile=/dev/null $worker_ip:stats ./$worker_ip.stats
-    check_exit_code "ERROR: Failed to fetch results from $worker_ip"
-    # TODO: this will overwrite results if we have more  than 1 worker, investigate this case
-    python3 speed-tester-analysis.py ./$worker_ip.stats $worker_ip results_summary.stats
+    # get the benchmarks for each node (some servers are faster)
+    . ./$worker_ip/benchmarks
+    # TODO: this will overwrite results if we have more than 1 worker, investigate this case
+    if [[ "$RUN_MODE" -eq "0" ]]
+    then
+        # fetch pktgen stats 
+        fetch_files $worker_key_file $worker_ip pktgen_stats
+        python3 pktgen-analysis.py ./$worker_ip.pktgen_stats $worker_ip pktgen_summary.stats $AVG_PKTGEN_SPEED
+        check_exit_code "Failed to parse Pktgen stats"
+        # fetch speed_tester stats
+        fetch_files $worker_key_file $worker_ip speed_stats
+        python3 speed-tester-analysis.py ./$worker_ip.speed_stats $worker_ip speed_summary.stats $AVG_SPEED_TESTER_SPEED
+        check_exit_code "Failed to parse Speed Tester stats"
+    else
+        # only fetch speed tester stats if mode is not 0
+        fetch_files $worker_key_file $worker_ip speed_stats
+        python3 speed-tester-analysis.py ./$worker_ip.speed_stats $worker_ip speed_summary.stats $AVG_SPEED_TESTER_SPEED
+        check_exit_code "Failed to parse Speed Tester stats"
+    fi
     check_exit_code "ERROR: Failed to analyze results from $worker_ip"
 done