-
Notifications
You must be signed in to change notification settings - Fork 16
AttachVolume.Attach failed for volume invalid character 'j' looking for beginning of value #50
Comments
Logs from ovirt-flexvolume-driver
Logs from volume provisioner
|
I need some more logs from the node but I think I know what's going on, the flexdriver on the node does not send the vm name it should attach the disk too. Can you check the driver conf file on node 091: It should have a conf item If its not there then something in the deployment was missing. How did you deploy it, using the apb? I guess that the best way is to auto detect it from a node lablel and get it as a env var for the deploying container to copy it over to the conf. |
Here are the contents of that file:
I deployed it the manual way, with the following command:
The environment is tightly locked down, so this was the easiest way to get started. I can try redeploy using the apb, if you believe that will help. I can provide more logs, but it seems like you were correct already. |
I'll make the deployment better and the discovery of ovirtVmName seamless. Is it working now that you set it? |
Does seem to be working.
This is the error that stands out in the logs on that server
File looks like this now
Would i need to reload any of the ovirt pods? |
Additional error in the oc describe po
|
No you don't need to refresh the pods, it should picked that up. Is nbschv091.telkom.co.za matches the name of the vm in ovirt? |
Yes. |
Obviously that annoying 'j' character is coming from the error while invoking the ovirt-api so first
|
|
curl -k -u ovirt-openshift-extensions@internal:XXXX https://nbschv109.telkom.co.za/ovirt-engine/api/disks Is rather large, so here is the file. |
Here is the journal of today for the node: nbschv091.telkom.co.za |
engine.logs |
@righteousjester Are you still getting this? I can't reproduce this with openshift 3.9.0 |
I am going to redeploy everything again to see if the error has cleared. BTW, I tried the automatic deploy and it fails on this step
I am running as root, but it looks like this path Will redeploy it following the manual steps for now |
I think you bumped into #30 Is this a problem with the README that is not updated? |
The README file looks fine.
I don't think it is pulling through the $HOME. when running:
But can leave that for another discussion. |
Still getting the error. Logs from pod
Here is the deployment of the ovirt extension
If I look at the logs on that node 095, I see the errors:
Will attach the logs from that node for only a day ago, using |
I don't see the flexdriver in action in the log at all. Could it be that the kubelet is running in a container? |
I'm having similar issues -- pv and pvc is creating, I see the disk in ovirt, but it doesn't attach to my node vm. Also, I've had no luck getting apb to work at all. I'll try to catch you on irc -- I'd really like to get this working. I'm currently using ovirt 4.2.5 and origin 3.10. |
How are you guys deploying the nodes? what's the output of |
My nodes are running on regular (not atomic) CentOS 7 VMs. |
@righteousjester This is working for me now. I do see an invalid character error, but then the volume mount succeeds. |
Every debugging attempt should include the logs of the kube-controller-manager pod |
I'm going to close this, I think this is solved from v0.3.2 onward, probably even earlier. |
Receiving the following error when a pod tries to attach to a PV.
PV is created successfully
And I can confirm I can see the Disk in RHV.
The pod has be set to always start on the same node: nbschv091.telkom.co.za
I am picking on the docker-registry pod
nbschv091.telkom.co.za is one of our nodes selected to host Openshift infrastructure components
Here are the logs from that server when that pod starts up
Layout of the ovirt pods
nbschv094 to nbschv096 are application nodes.
The text was updated successfully, but these errors were encountered: