Skip to content
This repository has been archived by the owner on Nov 9, 2020. It is now read-only.

[Tenant] Inconsistent docker volume ls output in case of orphaned VMs #990

Closed
ashahi1 opened this issue Mar 3, 2017 · 5 comments
Closed
Assignees
Milestone

Comments

@ashahi1
Copy link
Contributor

ashahi1 commented Mar 3, 2017

Do not see consistent output for 'docker volume ls' command across vms after _DEFAULT tenant is deleted.

Setup:
One ESX,
Two VMs (TestVM1-125 and TestVM2-125)
VSAN datastore.
Both VMs belong to same vsan datastore and same esx.

Steps:

  1. Created a docker volume named testVolVsan from vm2 (TestVM2-125).
root@sc-rdops-vm02-dhcp-52-237:~# docker volume create --driver=vsphere --name=testVolVsan -o size=200mb
testVolVsan
root@sc-rdops-vm02-dhcp-52-237:~# docker volume ls
DRIVER              VOLUME NAME
vsphere             testVolVsan@vsanDatastore
vsphere             vol1@sharedVmfs-0
vsphere             vol2@sharedVmfs-0

Docker volume ls command from VM1

root@sc-rdops-vm02-dhcp-52-237:~# docker volume ls
DRIVER              VOLUME NAME
vsphere             testVolVsan@vsanDatastore
vsphere             vol1@sharedVmfs-0
vsphere             vol2@sharedVmfs-0
  1. Admin cli command output listing volumes and default tenant .
[root@sc-rdops-vm18-dhcp-55-125:~] /usr/lib/vmware/vmdkops/bin/vmdkops_admin.py ls
Volume       Datastore      Created By VM  Created                   Attached To VM (name/uuid)  Policy          Capacity  Used   Disk Format  Filesystem Type  Access      Attach As
-----------  -------------  -------------  ------------------------  --------------------------  --------------  --------  -----  -----------  ---------------  ----------  ----------------------
vol1         sharedVmfs-0   ubuntu-VM0.0   Fri Mar  3 19:21:15 2017  detached                    N/A             10GB      145MB  thin         ext4             read-write  independent_persistent
vol2         sharedVmfs-0   ubuntu-VM1.0   Fri Mar  3 19:21:34 2017  detached                    N/A             10GB      145MB  thin         ext4             read-write  independent_persistent
testVolVsan  vsanDatastore  TestVM2-125    Fri Mar  3 19:55:00 2017  detached                    [VSAN default]  200MB     96MB   thin         ext4             read-write  independent_persistent

[root@sc-rdops-vm18-dhcp-55-125:~] /usr/lib/vmware/vmdkops/bin/vmdkops_admin.py tenant ls
Uuid                                  Name      Description               Default_datastore  VM_list
------------------------------------  --------  ------------------------  -----------------  -------
11111111-1111-1111-1111-111111111111  _DEFAULT  This is a default tenant
  1. Then deleted the default tenant. Do not see any volumes or default tenant
[root@sc-rdops-vm18-dhcp-55-125:~] /usr/lib/vmware/vmdkops/bin/vmdkops_admin.py tenant rm --name=_DEFAULT
tenant rm succeeded
[root@sc-rdops-vm18-dhcp-55-125:~] /usr/lib/vmware/vmdkops/bin/vmdkops_admin.py tenant ls
Uuid  Name  Description  Default_datastore  VM_list
----  ----  -----------  -----------------  -------


[root@sc-rdops-vm18-dhcp-55-125:~] /usr/lib/vmware/vmdkops/bin/vmdkops_admin.py ls
Volume  Datastore  Created By VM  Created  Attached To VM (name/uuid)  Policy  Capacity  Used  Disk Format  Filesystem Type  Access  Attach As
------  ---------  -------------  -------  --------------------------  ------  --------  ----  -----------  ---------------  ------  ---------

  1. Do not see same output for for docker volume ls command.

'docker volume ls' command from VM1

root@sc-rdops-vm02-dhcp-52-237:~# docker volume ls
list vsphere: VolumeDriver.List: VM TestVM1-125 does not belong to any tenant
DRIVER              VOLUME NAME

'docker volume ls' command from VM2

root@sc-rdops-vm02-dhcp-52-237:~# docker volume ls
list vsphere: VolumeDriver.List: VM TestVM2-125 does not belong to any tenant
DRIVER              VOLUME NAME
vsphere             testVolVsan
vsphere             testVolVsan@vsanDatastore
root@sc-rdops-vm02-dhcp-52-237:~#

Logs:
docker-volume-vsphere-VM2.txt
vmdk_ops.txt
docker-volume-vsphere-VM1.txt

@ashahi1 ashahi1 added this to the 0.13 milestone Mar 3, 2017
@ashahi1
Copy link
Contributor Author

ashahi1 commented Mar 3, 2017

CC / @pshahzeb

@ashahi1
Copy link
Contributor Author

ashahi1 commented Mar 3, 2017

Also, after deleting the tenant, we see two volumes being listed on the vm if we do 'docker volume ls' - one short name and another long name.

'docker volume ls' command from VM2

root@sc-rdops-vm02-dhcp-52-237:~# docker volume ls
list vsphere: VolumeDriver.List: VM TestVM2-125 does not belong to any tenant
DRIVER              VOLUME NAME
vsphere             testVolVsan  <<<<<<<<<<<<<<<<<<<<<<< 1
vsphere             testVolVsan@vsanDatastore <<<<<<<<<<<<<<<< 2
root@sc-rdops-vm02-dhcp-52-237:~#

@msterin
Copy link
Contributor

msterin commented Mar 7, 2017

After deleting _DEFAULT tenant the following should happen by design:

  • All Vms which do not belong to any tenant (i.e. the ones formerly in _DEFAULT) should be denied access to anything. 'docker volume ls' should say warning: VM XXX does not belong to any tenant and show empty list. docker volume create should fail
    //CC @lipingxue

@msterin
Copy link
Contributor

msterin commented Mar 14, 2017

@pshahzeb can you decipher the title and be clear on what's wrong compared to what was expected ? Currently it requires reading the lots of text just to figure it out.

it looks to me you are saying two different things:

  1. After removing _DEFAULT tenant, volumes dissapear form 'vmdkops_admin ls' command. , This is a clear bug, admin command should show all volumes. If the volumes do not belong to ANY tenant. we still should show them , I'd suggest marking tenant name (i.e. folder name) with *. This code does not exist (to mark stuff with *) but all in all this is a bug
  2. After removing _DEFAULT tenant. docker volume ls command still shows some volumes.
    You need to restart docker service , most likely it caches the list. If after docker restart all is as expected, this is most likely a docker bug. We do need to check what do we return to tdocker on List() when volumes are hidden by access control.

@pshahzeb pshahzeb changed the title [Tenanat] After deleting default tenant, not seeing same output for 'docker volume ls' across VMs. [Tenant] Inconsistent docker volume ls output in case of orphaned VMs Mar 17, 2017
@pshahzeb
Copy link
Contributor

@msterin Yes. The two things are apt as pointed.
This issue focuses on the second point.
The first part is being handled as part of #1043

pshahzeb pushed a commit that referenced this issue Mar 21, 2017
1. Update the detach disk flow to detach a disk irrespective of tenant and privilege.
   Update the attach disk flow to allow unique named volumes to be attached to a VM
   to ensure no edge cases exists of duplicate entries of disks attached to a VM

2. Send empty list as output to docker volume ls command to forcefully make docker
   print emtpy list and not the cached volume names

3. Minor print to indicate negative test case and avoid CI output confusion

Fixes #990 #1045
pshahzeb pushed a commit that referenced this issue Mar 21, 2017
1. Update the detach disk flow to detach a disk irrespective of tenant and privilege.

2. Send empty list as output to docker volume ls command to forcefully make docker
   print emtpy list and not the cached volume names

3. Minor print to indicate negative test case and avoid CI output confusion

Fixes #990 #1045
pshahzeb pushed a commit that referenced this issue Mar 30, 2017
1. Disallowing tenant-vm membership change when volumes are mounted and in use by VMs
2. Send empty list as output to docker volume ls command to forcefully make docker
   print emtpy list and not the cached volume names

3. Minor print to indicate negative test case and avoid CI output confusion

Fixes #990 #1045
pshahzeb pushed a commit that referenced this issue Mar 31, 2017
* Handling tenancy related edge cases

1. Disallowing tenant-vm membership change when volumes are mounted and in use by VMs
2. Send empty list as output to docker volume ls command to forcefully make docker
   print emtpy list and not the cached volume names

3. Minor print to indicate negative test case and avoid CI output confusion

Fixes #990 #1045
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
Development

No branches or pull requests

3 participants