* Admin Can Specify in Which GCE Availability Zone(s) a PV Shall Be Created An admin wants to specify in which GCE availability zone(s) users may create persistent volumes using dynamic provisioning. That's why the admin can now configure in StorageClass object a comma separated list of zones. Dynamically created PVs for PVCs that use the StorageClass are created in one of the configured zones. * Admin Can Specify in Which AWS Availability Zone(s) a PV Shall Be Created An admin wants to specify in which AWS availability zone(s) users may create persistent volumes using dynamic provisioning. That's why the admin can now configure in StorageClass object a comma separated list of zones. Dynamically created PVs for PVCs that use the StorageClass are created in one of the configured zones. * move hardPodAffinitySymmetricWeight to scheduler policy config * Added Bind method to Scheduler Extender - only one extender can support the bind method - if an extender supports bind, scheduler delegates the pod binding to the extender * examples/podsecuritypolicy/rbac: allow to use projected volumes in restricted PSP. * fix typo * SPBM policy ID support in vsphere cloud provider * fix the invalid link * DeamonSet-DaemonSet * Update GlusterFS examples readme. Signed-off-by: Humble Chirammal <hchiramm@redhat.com> * fix some typo in example/volumes * Fix spelling in example/spark * Correct spelling in quobyte * Support custom domains in the cockroachdb example's init container This switches from using v0.1 of the peer-finder image to a version that includes https://github.com/kubernetes/contrib/pull/2013 While I'm here, switch the version of cockroachdb from 1.0 to 1.0.1 * Update docs/ URLs to point to proper locations * Adds --insecure to cockroachdb client command Cockroach errors out when using said command: ```shell ▶ kubectl run -it --rm cockroach-client --image=cockroachdb/cockroach --restart=Never --command -- ./cockroach sql --host cockroachdb-public Waiting for pod default/cockroach-client to be running, status is Pending, pod ready: false Waiting for pod default/cockroach-client to be running, status is Pending, pod ready: false Waiting for pod default/cockroach-client to be running, status is Pending, pod ready: false If you don't see a command prompt, try pressing enter. Error attaching, falling back to logs: unable to upgrade connection: container cockroach-client not found in pod cockroach-client_default Error: problem using security settings, did you mean to use --insecure?: problem with CA certificate: not found Failed running "sql" Waiting for pod default/cockroach-client to terminate, status is Running pod "cockroach-client" deleted ``` This PR updates the README.md to include --insecure in the client command * Add StorageOS volume plugin * examples/volumes/flexvolume/nfs: check for jq and simplify quoting. * Remove broken getvolumename and pass PV or volume name to attach call * Remove controller node plugin driver dependency for non-attachable flex volume drivers (Ex: NFS). * Add `imageFeatures` parameter for RBD volume plugin, which is used to customize RBD image format 2 features. Update RBD docs in examples/persistent-volume-provisioning/README.md. * Only `layering` RBD image format 2 feature should be supported for now. * Formatted Dockerfile to be cleaner and precise * Update docs for user-guide * Make the Quota creation optional * Remove duplicated line from ceph-secret-admin.yaml * Update CockroachDB tag to v1.0.3 * Correct the comment in PSP examples. * Update wordpress to 4.8.0 * Cassandra example, use nodetool drain in preStop * Add termination gracePeriod * Use buildozer to remove deprecated automanaged tags * Use buildozer to delete licenses() rules except under third_party/ * NR Infrastructure agent example daemonset Copy of previous newrelic example, then modified to use the new agent "newrelic-infra" instead of "nrsysmond". Also maps all of host node's root fs into /host in the container (ro, but still exposes underlying node info into a container). Updates to README * Reduce one time url direction Reduce one time url direction * update to rbac v1 in yaml file * Replicate the persistent volume label admission plugin in a controller in the cloud-controller-manager * update related files * Paramaterize stickyMaxAgeMinutes for service in API * Update example to CockroachDB v1.0.5 * Remove storage-class annotations in examples * PodSecurityPolicy.allowedCapabilities: add support for using * to allow to request any capabilities. Also modify "privileged" PSP to use it and allow privileged users to use any capabilities. * Add examples pods to demonstrate CPU manager. * Tag broken examples test as manual * bazel: use autogenerated all-srcs rules instead of manually-curated sources rules * Update CockroachDB tag to v1.1.0 * update BUILD files * pkg/api/legacyscheme: fixup imports * Update bazel * [examples.storage/minio] update deploy config version * Volunteer to help review examples I would like to do some code review for examples about how to run real applications with Kubernetes * examples/podsecuritypolicy/rbac: fix names in comments and sync with examples repository. * Update storageclass version to v1 in examples * pkg/apis/core: mechanical import fixes in dependencies * Use k8s.gcr.io vanity domain for container images * Update generated files * gcloud docker now auths k8s.gcr.io by default * -Add scheduler optimization options, short circuit all predicates if one predicate fails * Revert k8s.gcr.io vanity domain This reverts commit eba5b6092afcae27a7c925afea76b85d903e87a9. Fixes https://github.com/kubernetes/kubernetes/issues/57526 * Autogenerate BUILD files * Move scheduler code out of plugin directory. This moves plugin/pkg/scheduler to pkg/scheduler and plugin/cmd/kube-scheduler to cmd/kube-scheduler. Bulk of the work was done with gomvpkg, except for kube-scheduler main package. * Fix scheduler refs in BUILD files. Update references to moved scheduler code. * Switch to k8s.gcr.io vanity domain This is the 2nd attempt. The previous was reverted while we figured out the regional mirrors (oops). New plan: k8s.gcr.io is a read-only facade that auto-detects your source region (us, eu, or asia for now) and pulls from the closest. To publish an image, push k8s-staging.gcr.io and it will be synced to the regionals automatically (similar to today). For now the staging is an alias to gcr.io/google_containers (the legacy URL). When we move off of google-owned projects (working on it), then we just do a one-time sync, and change the google-internal config, and nobody outside should notice. We can, in parallel, change the auto-sync into a manual sync - send a PR to "promote" something from staging, and a bot activates it. Nice and visible, easy to keep track of. * Remove apiVersion from scheduler extender example configuration * Update examples to use PSPs from the policy API group. * fix all the typos across the project * Autogenerated: hack/update-bazel.sh * Modify PodSecurityPolicy admission plugin to additionally allow authorizing via "use" verb in policy API group. * fix todo: add validate method for &schedulerapi.Policy * examples/podsecuritypolicy: add owners. * Adding dummy and dummy-attachable example Flexvolume drivers; adding DaemonSet deployment example * Fix relative links in README |
||
---|---|---|
.. | ||
README.md | ||
javaweb-2.yaml | ||
javaweb.yaml | ||
workflow.png |
README.md
Java Web Application with Tomcat and Sidecar Container
The following document describes the deployment of a Java Web application using Tomcat. Instead of packaging war
file inside the Tomcat image or mount the war
as a volume, we use a sidecar container as war
file provider.
Prerequisites
https://github.com/kubernetes/kubernetes/blob/master/docs/user-guide/prereqs.md
Overview
This sidecar mode brings a new workflow for Java users:
As you can see, user can create a sample:v2
container as sidecar to "provide" war file to Tomcat by copying it to the shared emptyDir
volume. And Pod will make sure the two containers compose an "atomic" scheduling unit, which is perfect for this case. Thus, your application version management will be totally separated from web server management.
For example, if you are going to change the configurations of your Tomcat:
$ docker exec -it <tomcat_container_id> /bin/bash
# make some change, and then commit it to a new image
$ docker commit <tomcat_container_id> mytomcat:7.0-dev
Done! The new Tomcat image will not mess up with your sample.war
file. You can re-use your tomcat image with lots of different war container images for lots of different apps without having to build lots of different images.
Also this means that rolling out a new Tomcat to patch security or whatever else, doesn't require rebuilding N different images.
Why not put my sample.war
in a host dir and mount it to tomcat container?
You have to manage the volumes in this case, for example, when you restart or scale the pod on another node, your contents is not ready on that host.
Generally, we have to set up a distributed file system (NFS at least) volume to solve this (if we do not have GCE PD volume). But this is generally unnecessary.
How To Set this Up
In Kubernetes a Pod is the smallest deployable unit that can be created, scheduled, and managed. It's a collocated group of containers that share an IP and storage volume.
Here is the config javaweb.yaml for Java Web pod:
NOTE: you should define war
container first as it is the "provider".
apiVersion: v1
kind: Pod
metadata:
name: javaweb
spec:
containers:
- image: resouer/sample:v1
name: war
volumeMounts:
- mountPath: /app
name: app-volume
- image: resouer/mytomcat:7.0
name: tomcat
command: ["sh","-c","/root/apache-tomcat-7.0.42-v2/bin/start.sh"]
volumeMounts:
- mountPath: /root/apache-tomcat-7.0.42-v2/webapps
name: app-volume
ports:
- containerPort: 8080
hostPort: 8001
volumes:
- name: app-volume
emptyDir: {}
The only magic here is the resouer/sample:v1
image:
FROM busybox:latest
ADD sample.war sample.war
CMD "sh" "mv.sh"
And the contents of mv.sh
is:
cp /sample.war /app
tail -f /dev/null
Explanation
- 'war' container only contains the
war
file of your app - 'war' container's CMD tries to copy
sample.war
to theemptyDir
volume path - The last line of
tail -f
is just used to hold the container, as Replication Controller does not support one-off task - 'tomcat' container will load the
sample.war
from volume path
What's more, if you don't want to enclose a build-in mv.sh
script in the war
container, you can use Pod lifecycle handler to do the copy work, here's a example javaweb-2.yaml:
apiVersion: v1
kind: Pod
metadata:
name: javaweb-2
spec:
containers:
- image: resouer/sample:v2
name: war
lifecycle:
postStart:
exec:
command:
- "cp"
- "/sample.war"
- "/app"
volumeMounts:
- mountPath: /app
name: app-volume
- image: resouer/mytomcat:7.0
name: tomcat
command: ["sh","-c","/root/apache-tomcat-7.0.42-v2/bin/start.sh"]
volumeMounts:
- mountPath: /root/apache-tomcat-7.0.42-v2/webapps
name: app-volume
ports:
- containerPort: 8080
hostPort: 8001
volumes:
- name: app-volume
emptyDir: {}
And the resouer/sample:v2
Dockerfile is quite simple:
FROM busybox:latest
ADD sample.war sample.war
CMD "tail" "-f" "/dev/null"
Explanation
- 'war' container only contains the
war
file of your app - 'war' container's CMD uses
tail -f
to hold the container, nothing more - The
postStart
lifecycle handler will docp
after thewar
container is started - Again 'tomcat' container will load the
sample.war
from volume path
Done! Now your war
container contains nothing except sample.war
, clean enough.
Test It Out
Create the Java web pod:
$ kubectl create -f examples/javaweb-tomcat-sidecar/javaweb-2.yaml
Check status of the pod:
$ kubectl get -w po
NAME READY STATUS RESTARTS AGE
javaweb-2 2/2 Running 0 7s
Wait for the status to 2/2
and Running
. Then you can visit "Hello, World" page on http://localhost:8001/sample/index.html
You can also test javaweb.yaml
in the same way.
Delete Resources
All resources created in this application can be deleted:
$ kubectl delete -f examples/javaweb-tomcat-sidecar/javaweb-2.yaml