ibm / cloudland

Cloudland, a light weight infrastructure as a service project, plus OpenShift 4 platform as a service deployment engine, is a system framework to manage VM instances, software defined networks (SDN), volumes ..., It can handle over 10 thousand hypervisors in one cluster so it can be a base of large scale public cloud. More over, with multi-tenant and OpenShift 4 cluster deployments on demand, it can be a straightforward alternative for private cloud or as a hyper converged infrastructure (HCI) solution.

CloudLand

Cloudland, a light weight infrastructure as a service project, plus OpenShift 4 platform as a service deployment engine, is a system framework to manage VM instances, software defined networks (SDN), volumes ..., It can handle over 10 thousand hypervisors in one cluster so it can be a base of large scale public cloud. More over, with multi-tenant and OpenShift 4 cluster deployments on demand, it can be a straightforward alternative for private cloud or as a hyper converged infrastructure (HCI) solution.

Cloudland's main distinguishing features are:

• Able to deploy OpenShift 4 cluster on demand per tenant
• Compatibility with Openstack API (TBD)
• Light weight, no tons of components
• Flat learning curve for both developers and operators
• Excellent performance for inside messages delivery
• Based on HPC architecture, so super scalable
• Self auto fail recovery and stable
• Easy to customize to implement your own feature

Architecture overview

To support ultra-large scale, the hypervisors are organized into a tree hierarchy, the agents (scia are launched on demand)

For more information, see the Introduction

Installation

Support two ways to install cloudland

1. Install cloudland with rpm package in a quick starter
   • Refer to Installation to get more details
2. Build source code and install cloudland from end to end
   • Refer to Build and Installation to get more informations

User Guide

• Launch Virtual Machine
• Create Openshift Cluster
• Security Groups

For more usage, refer to User Manual

Reporting Issues

If you encounter any problem with this package, please open an issue tracker to us

Contributing

Refer to CONTRIBUTING

License

Apache License 2.0, see LICENSE.

Visit doc for full documentation and guide.

cloudland's Issues

Install integration with noVNC into allinone.sh

1. copy noVNC static files into web/clui/public
2. install and configure libvirt-console-proxy
3. test on the installation

Implement libvirt console resolver in handler

1. Need to authenticate with JWT token
2. Need to return instance vnc IP & port

dirty data generated if no ip can be used when creating one new interface

Could you given some introduction of config for allinone.sh file?

Could you given some introduction of config for allinone.sh file?

instances_patch:37:14: executing "instances_patch" at <eq .Type "internal">: error calling eq: invalid type for comparison

High availability VM

Initialize the code base

Submit the code base to start the contribution.

Recover routers after reboot

Implement console redirect url with token handler

return a status code 301 with location of console url with token

Need to apply security group

Add TLS for web clui

Need a way to recover routers when both routers are down

During maintenance, it is possible that both virtual routers on different hypervisors are gone if both hypervisors are down, so it needs a way to recover it on another hypervisor or migrate the router beforehand.

Fail to update security group for existing interface

Compatible Openstack API v3

Enable paginations in UI

[Task] subnet update support

Scheduler improvement

Provide OCP console URL link and kubeadm password after an OCP cluster installed

Provide the OCP console link and kubeadm password on the UI and let user open the console directly.

network not available after run allinone.sh

after run allinone.sh using cland user, network is down
And I tried to restart network, it works few seconds, after that, down again.

Hard to know which one is the expected one when creating floatingip

Can not login VM with pub key generated by putty

Subnet's hyper and peer should reflect the actual location reported by report_rc.sh

During maintenance, sometimes we need to copy the subnet config files to another machine. Therefore, the hyper and peer should change accordingly.

[Bug] the old instance is still available after migration, and 'start' action will start both instances.

UI improvement

Make the UI more beautiful, more convenient and reasonable to use.

Deployment problems summary

1.Is it necessary to run allinone.sh in cland user?
2.Selinux should be disabled
3.If the IP of hypervisor is 192.168.1.0/24, some unknown problems maybe happen in deploy process, eg: cloudland compile failed, clui compile failed etc. Haven't tried 172.16.20.0/24, same question may happen too?
4.After I changed IP of hypervisor from 192.168.1.0/24 to 192.168.168.0/24, most problems solved, but cloudland did not compile, need to compile it manually.Before compile, need to stop cloudlet.
5.The env param ADMIN_PASSWD I changed in allinone.sh, but it did not set as the default password of admin, is it ok?
6.cloudland project is too big, nearly 80M, is it ok? When git clone in bad network environment, may cause clone fail problem.
7.Suggest to transfer some code to Aliyun or some repo in China.

Add left navigation into all _new templates

Need to replace the first level of navigation with secondary level of navigation

Openshift cluster auto deployment

OpenShift VM hostname naming convention

<%cluster_name%>-master
<%cluster_name%>-worker
<%cluster_name%>-lb

Command execution in parallel on compute node

VM resize capability

VM console or vpc redirect

Specify designated LB external IP for an OCP cluster

Allow a user to specify a designated IP for LB.

[Bug] the new VM instance is not accessible after migration

Assign flavor to master and worker nodes

Allow users to select/specify flavor to master and worker nodes to be deployed.

Add drop down list to actions

In instances list view, change the delete button to a drop down list which is going to have migrate, delete, resize ...

org name is not passed to backend during creation

INSERT  INTO "organizations" ("created_at","updated_at","deleted_at","uuid","creater","owner","name","default_sg") VALUES ('2019-11-04 09:01:48','2019-11-04 09:01:48',NULL,'363b3f37-7e3b-43d4-b792-39ce17fa3c19',1,1,'',0) RETURNING "organizations"."id"

Suggest to add copy kvm image function

Suggest to add copy kvm image function
It will be easy to migrate, thx

User can select image for LB when provision an OCP cluster

A user can select different cloud image for LB when provision an OCP cluster.

The error should display after deletion failed

Now user can not see any error from web console after a failed deletion.

Multiple hypervisors and more IPs type issue

Tried to user the 3rd way to set my network, but Error: Failed to modify connection 'enp2s0': ipv4.dns: this property is not allowed for 'method=disabled' happened.
And check ip add found device br5000 not UP br5000: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default qlen 1000 link/ether 46:da:ef:2c:6d:3f brd ff:ff:ff:ff:ff:ff inet 192.168.168.202/24 brd 192.168.168.255 scope global noprefixroute br5000 valid_lft forever preferred_lft forever shown the network device start up failed.
My network config is TYPE="Ethernet" PROXY_METHOD="none" BROWSER_ONLY="no" BOOTPROTO="static" DEFROUTE="yes" IPV4_FAILURE_FATAL="no" IPV6INIT="yes" IPV6_AUTOCONF="yes" IPV6_DEFROUTE="yes" IPV6_FAILURE_FATAL="no" IPV6_ADDR_GEN_MODE="stable-privacy" NAME="enp2s0" DEVICE="enp2s0" ONBOOT="yes" IPADDR="192.168.168.202" PREFIX="24" DNS1="192.168.168.1" GATEWAY="192.168.168.1" IPV6_PRIVACY="no"

Production installation

router doesn't work well on virtual machine with floating ip

Set quota for all resources

Support multi-zone

One web UI working with multiple cloudland, and the subnets can connect across different zones.

[Bug] 'hyper' field is always in editable mode

No way to know if the migration occur or not?

Image upload function

Found multiple security groups can be default

Broken links within docs

Hey there,

I noticed you have references to documentation with broken links. It's possible this is porting to a website and therefore this isn't useful... but in case it's not, the .md is missing in a number of places. For example:

• Operations guide: at end of the article you link to the Deployment guide without .md
• Deployment: at the end of the article you link to Operations without .md

Separately, in the main README:

Visit doc for full documentation and guide.

The link for docs should be https://github.com/IBM/cloudland/tree/master/doc

Hypervisor management

Subnet with dnsmasq is not cleaned up when deleting

There are garbage bridges and config files with dnsmasq process not cleaned up.