- Mar 25, 2024
-
-
Nathan Brown authored
Add readthedocs config file to ensure documentation is built using the expected environment. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Change-Id: If30002f2d629fe0d7e742f0519e60928ea47d5ca
-
- Mar 20, 2024
-
-
Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: Ib1bf86af4c7e775830241c5949797fa6daf64173
-
- Mar 18, 2024
-
-
Sanjyot Vaidya authored
Modified license years in few files for easier testing Since updating license patch of correct years was submitted in 2024, we have to add 2024 year in few files. Fixes: 00bf142e ("docs: updating license years") Signed-off-by:
Sanjyot Vaidya <Sanjyot.Vaidya@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Change-Id: I193b5c878a9d8695b42578af835cf17a8db4e8ef
-
- Mar 14, 2024
-
-
Sanjyot Vaidya authored
updating license years based on copyright notices: missed one file in previous commit Fixes: 00bf142e ("docs: updating license years") Signed-off-by:
Sanjyot Vaidya <Sanjyot.Vaidya@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Change-Id: I4dcab17f8ca77b5417b7265fea44dbc1a127599e
-
- Mar 13, 2024
-
-
Sanjyot Vaidya authored
updating license years based on copyright notices Signed-off-by:
Sanjyot Vaidya <Sanjyot.Vaidya@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Change-Id: I4cd4b3bcddd1bbd147b839e17bcc4ff37a5f7610
-
- Mar 12, 2024
-
-
The docs incorrectly tell the user to set the override to true when it needs to be false for EC2/PF based deployments. Fixes: 359ff197 ("doc: provide the correct arguments for testpmd") Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Change-Id: I194e2334ff6f741ed8ccf88cbedfecd14c06efd7
-
- Mar 08, 2024
-
-
Sanjyot Vaidya authored
Add v24.03 release notes Update EC2 cluster requirements security roles Signed-off-by:
Sanjyot Vaidya <Sanjyot.Vaidya@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Change-Id: Ibae5090e3cb88655740b22e75c14ce726e56b1bd
-
- Mar 01, 2024
-
-
Update the release tag to 24.03-RC0. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: Ie0c5a6f0752957c5d6e490c878f434af709eaf5d
-
- Feb 29, 2024
-
-
Currently, if a dataplane interface has an active route, it can cause dpdk-devbind.py to fail. Inform the user of the prerequisite of removing any active routes from the dataplane interface. Also add a section to the troubleshooting guide for this error. Reported-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I12ad118d937c1a8ef415a16b3e8a528ea6f4bb90
-
PCIe address in example inventory file had a typo. Fixes: 3081e4f1 ("doc: add EC2 documentation") Reported-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Change-Id: I9153d73229df2fd4005ff428d639ebc11f46dd70
-
Most pages needed a few minor edits. The list of edits are: - Remove FQDN setup from EC2 setup, which is IP based. - Remove "Get NIC PCIe Addresses section" from the dpdk testpmd documentation - Mention that a freshly rebooted node will need to disable swap and create VFs - Update kernel paramaters to be consistent in FAQ. - Update Maintainers. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Change-Id: Ia98be10ed5e2ea6f57a80b9b932ec2e7da71bb56
-
The override enabling end users to set the directory that hosts self signed certificates caused the playbook to crash. The root cause was that the override was redefined in the playbook. However, when the override is set, it will keep its user provided value rather than taking the value set by the task. Instead, use a separate variable rather than redefine it. Fixes: c3565293 ("cluster: place self-signed certs in non-volatile storage") Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I53c7f85edb741a8281d7439ace575921d1b55360
-
The solution design document detailed how VFs were orchestrated and added to Pods. Since PFs/ENIs are now supported, this document was extended to also discuss how PF/ENIs were orchestrated and added to Pods. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Change-Id: I084a9f763d9dcc77fe8759c15cdfd1e87539852a
-
- Feb 26, 2024
-
-
Two updates to the inventory file: - use ansible_private_key_file, as specified in the docs - use <fqdn_or_ec2_ip> rather than <fqdn>, as EC2 deployment does not use FQDNs. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Change-Id: I07c61a76b4c2f9a8f1cc68ff4ff4b7124eae5136
-
- Feb 21, 2024
-
-
In both userguide and quickstart guide the place where the command for running the testpmd sample application is documented, it needs to show the correct command line parameters that are to be used. Signed-off-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Reviewed-by:
Nathan Brown <Nathan.Brown@arm.com> Change-Id: Idda8d3cf537c8e4666cbc0d6d34d4b3ab48a31a7
-
- Feb 20, 2024
-
-
Previously, an end user could specify an output directory that did not exist, causing the dpdk-testpmd playbook to fail. Instead, ensure the output directory exists before attempting to put file(s) into the directory. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I88404d2e52a47aed93970221cd0b1f8db5ebbb0b
-
We need to wait for Calico to be ready before additional pods can be successfully created. The previous logic looped around the output of kubectl get pods, waiting for the calico-node pods to be 'Ready'. Instead, utilize kubectl's rollout status command to wait until the calico-node daemonset is ready. This has an additional benefit of easily setting a timeout based on the value of a variable. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: Ib863f74176643e28d3779d140ae9e816b8197cc9
-
The first two FAQ questions were about the definition of the project. That information does not belong in the FAQ, and should be covered in detail in other pages, like Overview. Suggested-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I5b7eed2174200599d3e1cb16592dec1b062c8490
-
- Feb 14, 2024
-
-
This reverts commit 6887d6e3 Reason for revert: deploy_on_vfs needs to be correctly set in both controller and worker node. Signed-off-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Change-Id: Ia16a069e8800cff64dada22178ab3db2feb1a1a2
-
- Feb 12, 2024
-
-
The image in the sample application did not match the style or detail of the remaining images in the documentation. The image now adopts the consistent color scheme, style, and detail of the remaining images. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I1d423ec0a8be6fa135a7e6bbd2ec229651aebdd6
-
The SR-IOV CNI is still installed even when deploying on PFs during a lab based installation, as the end user may want to use VFs in the future. This was not clear from existing wording. Suggested-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I252b7e15133cff740600eb35cf9148460e3e8bdb
-
- Jan 26, 2024
-
-
Improve verbiage of security disclaimer. Suggested-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Change-Id: I79e86a3652026ed6d35094fd2cc329ef32b30915
-
- Jan 23, 2024
-
-
Sanjyot Vaidya authored
added note for Ansible timeout Signed-off-by:
Sanjyot Vaidya <Sanjyot.Vaidya@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: Ia5338957bcee25985f395fbb4842b4dbda2e60f7
-
- Jan 18, 2024
-
-
Quickstart guide requires 4 isolated CPUs, while User Guide can support a 4 core worker machine with 3 isolated CPUs. Add this note to the isolcpus section to ensure the end user isolates the correct number of CPUs. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I2a56347560017ed670b52585e43772bca433d14d
-
Previously, the FAQ held both answers to common questions and tips on debugging common errors/issues. Instead, separate out the debugging steps into a new troubleshooting page. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Change-Id: I9ed84955e875223f5727a0bec7b9e89b861a6c5d
-
- Jan 12, 2024
-
-
Nathan Brown authored
With dpdk-testpmd sample application and relevant document updates, we can safely remove the dpdk-l3fwd sample application. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I1afb86eec921fdf034ba7b97f23743b5e74f6186
-
Instead of forcing deploy on vfs to false move it in the code to add this parameter Signed-off-by:
Sanjyot Vaidya <Sanjyot.Vaidya@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: Id99d3e223e96c3493b7e62ea82b8ef086e55c6e7
-
- Jan 09, 2024
-
-
The python library for docker recently released version 7.0.0. This contains a feature to validate the format of an image before building/pushing it to a registry. This validation step has a bug, causing it to reject valid tags if they have a port. Therefore, pin the version of docker library to below 7.0.0 until the bug fix is released. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: I634aa765625e6c7ae068fccda8d6b68f1d77edf4
-
- Jan 05, 2024
-
-
Add 4 tests to demonstrate and showcase the HA feature of Kubernetes in multinode cluster setup. The 4 tests are: 1. Taint a node manually 2. Use IPTable rules to simulate a network partition 3. Kill the kubelet process 4. Shutdown a node Signed-off-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Change-Id: I0712089152956c4e51be02b3436fea5bdcf837d3
-
Documentation of high availability of single pod failure in Kubernetes. Tried methods : 1. Using a kubectl delete pod 2. Using kill process inside the pod. Signed-off-by:
Sanjyot Vaidya <Sanjyot.Vaidya@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Change-Id: Ifb5b11ddf16ac0707fb7ba48b380f8019411305e
-
- updates overview with EC2 details. - updates quickstart guide to support EC2 deployment. - updates inventory template file. - updates user guide to support EC2 deployment. - updates FAQ questions with EC2 details. - updates relevant images for EC2 deployment. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: Ie63a18c4ee80c5af53e276eea91cfa700f5d9eed
-
- Jan 04, 2024
-
-
When worker only has 4 CPUs, only two CPUs can be isolated, otherwise kubeadm will not initialize the cluster. Under normal behavior, this would only leave 2 CPUs for Pods to run on, which is not enough to run a sample application with two dedicated, exclusive CPUs. The resolution is to combine the following steps: 1. isolate CPUs 1-2, leaving cores 0,3 to run Linux, kubelet, containerd, etc. 2. set reservedSystemCPUs to just CPU 0. This leaves CPUs 1-3 available for pods to run on 3. static CPU manager policy The static CPU manager will provide exclusive CPUs starting with the lowest CPU IDs. In this case, it will be CPUs 1 and 2, which are isolated. The remaining pods will share CPU 3, along with kubelet, containerd, Linux, etc. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Change-Id: Ib34b41a3aa21315bfe9ddd73fe85ee7a0d8e9970
-
Kubernetes is updating the URL and GPG key locations to install Kubernetes packages. Update to the latest locations before the old locations disappear. The new location has changed the revision number/format. See [0] for more details. [0]: https://kubernetes.io/blog/2023/08/15/pkgs-k8s-io-introduction/ Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Change-Id: I11bb150c6d2afb52c1bb3bfd0e8a07b601b0bb92
-
- Dec 08, 2023
-
-
Add relevant DPDK testpmd documentation to quickstart and user guide. Remove DPDK l3fwd documentation from the documentation. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Reviewed-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Change-Id: I4c52f4a6259832f95219d0c518b507c90ead4f8b
-
- Dec 06, 2023
-
-
Aditya Ambadipudi authored
On AWS igb_uio is used. It's not in the list of drivers. So everytime a CRA cluster on AWS is created, the ansible scripts throw a warning saying the driver is not a valid driver. Signed-off-by:
Aditya Ambadipudi <aditya.ambadipudi@arm.com> Reviewed-by:
Dhruv Tripathi <dhruv.tripathi@arm.com> Reviewed-by:
Nathan Brown <nathan.brown@arm.com> Change-Id: Ide966fa580007018920ad1f8c7b8e81be2ac7477
-
- Nov 29, 2023
-
-
dpdk-testpmd can be used in the 5 tuple swap (5tswap) forwarding mode to easily establish bidirectional traffic in cloud VPCs. Use LTS versions of Ubuntu and DPDK within the container. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Lijian Zhang <lijian.zhang@arm.com> Reviewed-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Tested-by:
Sanjyot Vaidya <sanjyot.vaidya@arm.com> Change-Id: I66be81190c56fe854c5953961438d1733cc4963f
-
- Oct 18, 2023
-
-
dpdk-devbind.py will skip binding any PF/VF that has an entry in the routing table. It logs this to stderr, but still exits 0. This can cause the Ansible role to succeed even if the PF/VF is not bound to the dpdk driver. Instead, fail the playbook if devbind is unable to correctly bind the PF/VF to the dpdk driver. Fixes: 2cb12ca2 ("cluster: replace driverctl with dpdk-devbind") Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Tianyu Li <tianyu.li@arm.com> Change-Id: I93571695cdbbe84f866b251bf157344c171baa73
-
- Oct 11, 2023
-
-
Amazon Linux EC2 deployment uses host-device CNI, and not SR-IOV CNI. This skips SR-IOV CNI building and installation when deploying on AWS. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Tianyu Li <tianyu.li@arm.com> Change-Id: Iec8618312e9ec06e142a664ad669c32f17e32cba
-
- Oct 10, 2023
-
-
Add a role to install AWS VPC CNI into a cluster created on AWS instances. Make slight adjustments to existing roles to provide required environment to install the CNI. Two notable changes include: 1. ec2-net-utils package is not present. This conflicts with the AWS VPC CNI in how secondary ENI IP addresses are handled. 2. Ensure that iptables FORWARD defaults to ALLOW instead of DENY. This is required to allow packets to be forwarded through the default network namespace to/from pod network namespaces. Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Tianyu Li <tianyu.li@arm.com> Change-Id: I4afe04d5dd69b7b28df7325a068de4db12e53548
-
- Sep 19, 2023
-
-
The controller role held all the code that was needed to setup and initialize the Kubernetes control plane, including installing a CNI and additional CNF components, like Multus CNI. This separates the controller role into three roles: - controller (to setup K8s control plane and docker registry) - calico (to install Calico CNI) - cnf_components (to install Multus/SR-IOV CNI and SR-IOV Net. Device Plugin). Signed-off-by:
Nathan Brown <nathan.brown@arm.com> Reviewed-by:
Tianyu Li <tianyu.li@arm.com> Change-Id: I9ad6a808627dd2c219776c8b4fc4410148b1a940
-