Dataverse v4.16 on K8s
Pre-releaseThis release contains references to the upstream release 4.16 and will deploy it when used.
Hotfix Release
BEWARE: this is a hotfixed 4.16 release to fix an issue in Jenkins, see #128 You should use the latest release, currently being 4.18.1
Pre-Deploy Actions
No special action seems to be necessary regarding to the upstream release docs.
- When using
kubectl
to upgrade Dataverse after upgrading Solr, you need to remove theDeployment
first, as the selector labels have changed (and they are immutable):kubectl delete deployment dataverse
kubectl apply -k k8s/dataverse
- Depending on your deployment mechanisms, you might need to add the
/docroot
PVC yourself. See below for context. You might usekubectl apply -k k8s/dataverse
, but see above.
Post-Deploy Actions
- The core bundled
citation
metadata block has changed. If you are upgrading
from an older release, you will need to upload the new schema and reexport.
Please see metadata docs on how to do that. - Remember when you upgrade from pre-v4.15, you will have to perform a re-index.
- None the less, please take a look at the rest of upstream release notes, too.
New and changed functionality
Use more Kustomize.io!
As Kubernetes 1.16 is out, people should really adopt using Kustomize.io bundled with kubectl
since 1.14.
This project supports the usage and welcomes the relief of easily modifiable K8s objects. See that easy call in the README... 😉 In future releases, we will take this even further, see #91.
Default admin password changed, can provision from Secret
As of this release (4.16) you should be aware that the default
dataverseAdmin
password is no longer admin, but admin1. As
IQSS/dataverse-ansible does the same, let's be consistent. See #75 for details.
Persistance for dataverse logos, sitemap, etc
Since this release, Dataverse will save logos for dataverses, its sitemap
and other statefull information at /docroot
of the container. See #65.
A PersistentVolumeClaim
and volume mount has been added. For now this is
in ReadWriteOnce
mode. For using it with a scaled, multi-instance deployment,
that should be changed to ReadWriteMany
. See also #10.
Support for running any branch in K8s
This release adds initial support for running any branch in Kubernetes by using
local Kind clusters and Skaffold for building and deploying images.
For more information, see the development-usage docs.
General Notes
Please be aware that future enhancements to scripting and more used to deploy and configure Dataverse will not be added to the release. Thus docker images for this will not be updated.
This is especially important in terms of security patches for images. This might change in the future with scheduled builds from CI.
Please be aware, that tagged images are not automatically updated by Kubernetes when updated on Docker Hub (image pull policy).