Added How Murano installs Kubernetes Cluster
Added How Murano upgrades Kubernetes Cluster
Minor updates on Networking
Fixed broken links
Added detailed info on Calico
Change-Id: Ic1538b33fa79ba0bb772a442dfc7c0a5db3a2e97
Following sections has been addedd:
* Features
* Rolling Updates of Kubernetes application
* Interacting with Kubernetes Cluster deployed by Murano
Co-Authored-By: Nikolay Mahotkin <nmakhotkin@mirantis.com>
Change-Id: I7d7572d6c53c9f6f0165ff55fa4d0ed5abdedef9
- binaries and scripts updated to use calico networking
Co-Authored-By: ddovbii <ddovbii@mirantis.com>
Change-Id: I3187b9cc171aa15caa4912a665c642ab057b4eae
1. default_scripts/kube-apiserver parametres updates:
Replaced: address, ports with insecure-bind-address, insecure-port
Added: service-cluster-ip-range, admission-control, advertise-address
Depricated: portal_net
2. environ/apiserver parametres updates:
Updated: KUBE_API_ADDRESS, KUBE_API_PORTS
Added: KUBE_API_IP
Depricated: KUBE_API_ARGS
3. systemd/kube-apiserver.service parametres updates:
Added: KUBE_API_IP
Update: KUBE_SERVICE_ADDRESSES with KUBE_SERVICE
Depricated: KUBE_API_ARGS
4. default_scripts/kube-controller-manager updates:
Added: New params for KUBE_CONTROLLER_MANAGER_OPTS such as
service-cluster-ip-range, cluster-cidr, cluster-name
5. systemd/kube-controller-manager.service updates:
Added: new argumunets to start up command such as
KUBE_SERVICE, KUBE_NAME, KUBE_CIDR
Removed: KUBE_CONTROLLER_MANAGER_ARGS
6. kube-config updates with new
Added new argumunets to start up command such as
KUBE_SERVICE, KUBE_NAME, KUBE_CIDR
7. scripts/master-kube-setup.sh update:
Set Master IP value for apiserver and kube-apiserver scripts
Change-Id: I808ef5459e5fd5f1f5c916ddd3f94beab05e5c5d
Scenario:
1. Create murano environment
2. Create session for create environment.
3. Initialize k8s cluster parameters and add app to env.
4. Initialize k8s pod parameters and add app to env.
5. Deploy session.
7. Check env status and port availability on k8s master node.
8. Get k8s minions' ips and check that correct initial number
of them was created and k8s api port is available on them.
9. Run 'scaleNodesUp' action for k8s minions.
10. Check that number of minions was increased and
k8s api port is available on all of them
11. Run 'scaleNodesDown' action for k8s minions.
12. Check that number of minions was decreased and
k8s api port is available on all of them
Change-Id: I63bf61b6f9e5fd45b8e6bde0fab87da5826bedc6
This commit does not resolve a bug with docker0 due to the fact
that murano-agent is unable to execute script for reconfiguration
So, this patch reverts commit 6547024fa3.
Change-Id: Ic522d54b14ade7878ce2c6b5729c1e117c329551
For running k8s and docker autotests against this repo
we need to prepare and upload packages first.
This is a POC of script, it should be refactored and improved in
future, but for now we need to have some prototype somewhere, otherwise
it will be hardcoded in jenkins job.
Change-Id: I58c7965b275dd3a86de69f9e341c61edc04cc976
* Since update command is replaced by replace here
https://review.openstack.org/#/c/358661/
for patching the resource it needs to use
'kubectl patch <type> <name> --patch <patch>'
Change-Id: Ie7e5c4aea3b0cee519ae5ed81cff0dcc3ab67562
* docker0 interface in ubuntu had incorrect MTU value
* Older docker (< v1.1) automatically infer MTU value based
on host. But since 1.1 docker doesn't do this. It
sets MTU value 1500 which causes some network problems
(especially with SSL/TLS/HTTPS) and apps like
HTTPdSite and NginXSite couldn't clone the repo.
Closes-Bug: 1608836
Change-Id: I62482bce9ad4f45b7ea2d1d4db77d009c7191f7b
* According to
http://kubernetes.io/docs/user-guide/volumes/#types-of-volumes
and
a1b5325011
there is hostPath volume type instead of hostDir;
And there is no 'source' keyword anymore.
* Replacing 'kubectl update' on 'kubectl replace':
kubectl update is deprecated, need to use 'replace':
9b3d42c090
Closes-Bug: #1604766
Change-Id: I1029caeb2827fc9de271b457c7a61b043d0189cb
* Due to the service name length limitation
it needs to be cut to 24 characters.
Closes-Bug: #1613651
Change-Id: Id2d68346034c0697f3fb41a090201e8736f8deec
* If we have a k8s cluster installed and
then deploy 2 apps, each on separate pod,
then in logs we will see 2 endpoints for
2nd app. Then we will see 3 endpoints for
next pod/app deployment. The reason is searching
endpoints was done by whole k8s cluster but it
should rather be only by current pod.
Now it is fixed.
Closes-Bug: #1613631
Change-Id: I730016dcd55a922ac6c9ee20f9de500f7831f9c3
* latest etcd is enabled
* ntp tools added for time sync
* Readme updated for tmpfs workaround when 8Gb memory is detected
Change-Id: Ifdfb72902fad843baef1fcee3e5bc65b119bdde3
