Open magick93 opened 7 years ago
gofabric8 volumes
tries to run minishift ssh
to execute commands inside the VM (such as making directories and chmodding - wonder if something goes wrong with the gofabric8 volumes
command? Any chance of seeing its output?
BTW gofabric8 volumes
is only intended for minikube and minishift. For real clusters you need to either setup a Storage class or setup some PVs. https://docs.openshift.com/container-platform/3.4/install_config/persistent_storage/dynamically_provisioning_pvs.html
BTW gofabric8 volumes is only intended for minikube and minishift.
Why is that? Its so useful! While I can set them using a volume json file, it is so easy to be able to run one command.
Any chance of seeing its output?
# gofabric8 volumes
▄▄▄▄▄▄▄
███████
▄▄▄▄▄▄▄ ▄▄▄▄▄▄█
███████ ▀▀▀▀▀▀▀ ▄▄▄▄▄▄
▄▄▄▄▄▄▄ ██████
▄▄▄▄▄ ▄▄▄▄▄▄
▀▄▄▄▄▄ ▄▄▄ ▄▄▄
▀▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄ ▄▄▄▄▄▀
▀▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄█▄▄▀
▀▀▄█▄▄▄▄▄▄▄▄▄▄▄▄▄▀
▄█▄▄▄█▄▄▄█▄▄▄
▄█▄▄▄▄▄▄▄▄▄▄▄
▄▄█▄▄▄▄▄▄▄▄
▄▄▄▄▀▀▄█▄
▄▄▄▄▄▄▄▄▄▄▄▄▄▄▄
▀▀▀▀▀▀▀▀▀▀▀▀▀▀
PersistentVolume name priceinsight-postgresql will be created on host path /data/priceinsight/postgresql
Recreating PVC postgresql
Any idea how I can get control (run commands) on my server again?
the problem trying to implement gofabric8 volumes
on anything else than a single node cluster is there are a large number of different Volume providers; each with completely different custom configuration. So its hard to tool that OOTB.
The better approach is for kubernetes & openshift clusters to come pre-configured with a StorageClass so that PVCs automatically bind to the OOTB storage implementation for the cluster (e.g. gluster/ceph for openshift on premise or EBS for on AWS and google file system for GKE etc). Most kubernetes clusters you create now come with a StorageClass configured OOTB so there's no need for gofabric8 volumes
at all.
So really the fix for this is for your OpenShift cluster to be configured with a proper StorageClass using whatever persistence provider you want your cluster to use; then fabric8 can just reuse PVCs on it
Yes, I understand that there are situations where it does not make sense to
use gofabric8 volumes
but in the case of a single node cluster, it does.
Nonetheless, it would be nice if it degrades gracefully, or at leasts gives some kind of feedback, as opposed to breaking the ability to interact with the host.
On 12 April 2017 at 09:15, James Strachan notifications@github.com wrote:
the problem trying to implement gofabric8 volumes on anything else than a single node cluster is there are a large number of different Volume providers; each with completely different custom configuration. So its hard to tool that OOTB.
The better approach is for kubernetes & openshift clusters to come pre-configured with a StorageClass so that PVCs automatically bind to the OOTB storage implementation for the cluster (e.g. gluster/ceph for openshift on premise or EBS for on AWS and google file system for GKE etc). Most kubernetes clusters you create now come with a StorageClass configured OOTB so there's no need for gofabric8 volumes at all.
So really the fix for this is for your OpenShift cluster to be configured with a proper StorageClass using whatever persistence provider you want your cluster to use; then fabric8 can just reuse PVCs on it
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/fabric8io/gofabric8/issues/366#issuecomment-293494261, or mute the thread https://github.com/notifications/unsubscribe-auth/ACEpHb9cCbRag_aa1t0_vyOujGqaBdhBks5rvHoZgaJpZM4M6NsN .
This is going to sound strange, but it has happened twice in a row now.
I have setup a new server, the host is running Centos 7, and I install OpenShift v1.5.0-rc.0+49a4a7a using OpenShift-ansible.
I then install Fabric8.
I then run
gofabric8 volumes
. After running this I lose the ability to run basically all commands on the server. The only command that still works iscd
.For example: