## ## Copyright (c) 2010-2017 Intel Corporation ## ## Licensed under the Apache License, Version 2.0 (the "License"); ## you may not use this file except in compliance with the License. ## You may obtain a copy of the License at ## ## http://www.apache.org/licenses/LICENSE-2.0 ## ## Unless required by applicable law or agreed to in writing, software ## distributed under the License is distributed on an "AS IS" BASIS, ## WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. ## See the License for the specific language governing permissions and ## limitations under the License. ## rapid (Rapid Automated Performance Indication for Dataplane) ************************************************************ rapid is a set of files offering an easy way to do a sanity check of the dataplane performance of an OpenStack environment. Copy the files in a directory on a machine that can run the OpenStack CLI commands and that can reach the OpenStack public network. Also create a qcow2 image in the same directory with the following characteristics: * Name of the qcow2 file should be: rapidVM.qcow2 This default name can be overruled on the rapid command line (--image_file) * Should have DPDK and PROX installed. PROX should be in /root/prox/ directory * Image should have cloud-init installed * /mnt/huge should exist to support a command that is executed at startup of the VM: 'mount -t hugetlbfs nodev /mnt/huge' * Compile prox with 'make crc=soft'. This is a workaround for some cases where the crc calculation offload is not working as expected. * Compile dpdk to support AESN-NI Multi Buffer Crypto Poll Mode Driver: http://dpdk.org/doc/guides/cryptodevs/aesni_mb.html Source the openrc file of the OpenStack environment so that the OpenStack CLI commands can be run: # source openrc Now you can run the createrapid.py file. Use help for more info on the usage: # ./createrapid.py --help createrapid.py will use the OpenStack CLI to create the flavor, key-pair, network, image, stack, ... It will create a .env file containing all info that will be used by runrapid.py to actually run the tests. Logging can be found in the CREATE.log file Now you can run the runrapid.py file. Use help for more info on the usage: # ./runrapid.py --help The script will connect to all VMs that have been instantiated and it will launch PROX in all VMs. This will be done through the floating IP assigned to the VMs. You need to make sure that floating IPs are working on your OpenStack deployment. Once that is done it will connect to the PROX tcp socket and start sending commands to run the actual test. It will print test results on the screen while running. The actual test that is running is decribed in .test. Notes about prox_user_data.sh script: - The script contains commands that will be executed using cloud-init at startup of the VMs. - The script also assumes some specific DPDK directory and tools which might change over different DPDK release. This release has been tested with DPDK-17.02. - huge pages are allocated for DPDK on node 0 (hard-coded) in the VM. Note on using SRIOV ports: Before running createrapid, it is best to already create the network, subnet and ports manually by entering the following commands (change the parameters to your needs): openstack network create --share --external --provider-network-type flat --provider-physical-network physnet2 fast-network openstack subnet create --network fast-network --subnet-range 20.20.20.0/24 --gateway none fast-subnet openstack port create --network fast-network --vnic-type direct --fixed-ip subnet=fast-subnet Port1 openstack port create --network fast-network --vnic-type direct --fixed-ip subnet=fast-subnet Port2 openstack port create --network fast-network --vnic-type direct --fixed-ip subnet=fast-subnet Port3 Make sure to use the network and subnet in the createrapid parameters list. Port1, Port2 and Port3 are being used in the yaml files. An example of the env file generated by createrapid.py can be found below. Note that this file can be created manually in case the stack is created in a different way (not using the createrapid.py). This can be useful in case you are not using OpenStack as a VIM or when using special configurations that cannot be achieved using createrapid.py. Only the [VMx] sections are used as input for runrapid.py. [DEFAULT] admin_ip = none [VM1] admin_ip = 192.168.4.130 dp_ip = 10.10.10.6 dp_mac = fa:16:3e:3c:1e:12 [VM2] admin_ip = 192.168.4.140 dp_ip = 10.10.10.9 dp_mac = fa:16:3e:2a:00:5d [VM3] admin_ip = 192.168.4.138 dp_ip = 10.10.10.11 dp_mac = fa:16:3e:ae:fa:86 [OpenStack] stack = rapid yaml = 3VMrapid.yaml key = prox flavor = prox_flavor image = rapidVM image_file = rapidVM.qcow2 dataplane_network = dataplane-network subnet = dpdk-subnet subnet_cidr = 10.10.10.0/24 internal_network = admin_internal_net floating_network = admin_floating_net [rapid] loglevel = DEBUG version = 17.10.25 total_number_of_vms = 3