Cleanup of TODO and rename of python tests

- removed useless TODOs. The majority of them are resolved,
and for those not resolved new issued are created (e.g. #205 #206)
- renamed tests_*.py to *_test.py to make consistent with go tests,
which are always named *_test.go (enforced by 'go test')

Makefile

@@ -200,7 +200,7 @@ ECHO = $(DEBUG) echo
 test-esx:
 	$(TAR) cz --no-recursion $(ESX_SRC)/*.py | $(SSH) root@$(ESX) "cd /tmp; $(TAR) xz"
 	$(ECHO) Running unit tests for vmdk-opsd python code on $(ESX)...
-	$(SSH) root@$(ESX) "python /tmp/$(ESX_SRC)/tests_vmdk_ops.py"
+	$(SSH) root@$(ESX) "python /tmp/$(ESX_SRC)/vmdk_ops_test.py"
 	$(SSH) root@$(ESX) rm -rf /tmp/$(ESX_SRC)
 
 testremote: test-esx test-vm 

plugin.go

@@ -1,43 +1,14 @@
 package main
 
 //
-// A VMDK Docker Data Volume plugin
-// (for now just an outline)
+// VMWare VMDK Docker Data Volume plugin.
 //
-// It checks that it can connect to ESX via vSocket , and then
-// fakes the successful ops by returning success and printing a messafe,
-// without doing anything useful (useful stuff is work in progress)
+// Provide suport for --driver=vmdk in Docker, when Docker VM is running under ESX.
 //
-// TBD: add check for the guest communicates from ROOT - check the doc wrt security
-// TBD: check we are running inside ESX and tools installed - in CODE , not make
-
-// TODO:
-//  convert all to return error() and only response to docker to return msg
-//
-// make sure msg and err is properly initialized everywhere (and check in go if it's needed)
+// This code received requests from Docker Engine and requests/coordinates
+// related VMDK operations with vmdk-opsd service running on enclosing host ESX 
+// (see ./vmdkops-esxsrv) 
 ///
-//TODO :
-// Potentially: add unit test , fully contained on 1 Linux machine:
-//- make server code actually create a loop block device and do a bind mount for volume test
-//- hardcode location for testing (/var/vmware/dvolplug/volumes/vol-name
-//	fallocate -d -l <length> volume.img
-//	losetup -v /dev/loop$(id) -f volume.img
-//	mkfs.ext4 /dev/loop$(id)
-//	mount /dev/lopp$(id) /mnt/loop$(id) # mount all here. Or skip
-//	mount -o bind # bind mount where Docker asks
-//	Good refs: https://www.suse.com/communities/blog/accessing-file-systems-disk-block-image-files/
-//)
-// for actual mounts:
-//- add actual create/mount code instead of prints
-//
-// TODO: add volumes tracking per the following docker spec:
-//
-// multiple containers on the same docker engine may use the same vmdk volume
-// thus we need to track the volumes already attached, and just do bind mount for them
-// Also it means we need to serialize all ops with mutex
-
-// We also need to track volumes attached and mounted to save on this ops if requested
-// On start, we need to list vmdks attached to the VM and polulate list of volumes from it
 
 import (
 	//	"encoding/json"
@@ -55,7 +26,7 @@ const (
 )
 
 type vmdkDriver struct {
-	m       *sync.Mutex // create() serialization
+	m       *sync.Mutex // create() serialization - FFU
 	mockEsx bool
 	ops     vmdkops.VmdkOps
 }
@@ -68,7 +39,6 @@ func newVmdkDriver(mockEsx bool) vmdkDriver {
 		vmdkCmd = vmdkops.VmdkCmd{}
 	}
 	d := vmdkDriver{
-		// TODO: volumes map(string)volinfo, (name->uuid/refcount/creationtime)
 		m:       &sync.Mutex{},
 		mockEsx: mockEsx,
 		ops:     vmdkops.VmdkOps{Cmd: vmdkCmd},
@@ -112,9 +82,10 @@ func (d vmdkDriver) mountVolume(r volume.Request, path string) error {
 		return nil
 	}
 
-	// TODO: refcount  if the volume is already mounted (for other container) and
+	// TODO: Issue #28
+	// - refcount  if the volume is already mounted (for other container) and
 	// just return volume.Response{Mountpoint: m} in this case
-	// TODO: save info abouf voliume mount , in memory
+	// -  save info abouf voliume mount , in memory
 	// d.volumes[m] = &volumeName{name: r.Name, connections: 1}
 	if err := d.ops.Attach(r.Name, r.Options); err != nil {
 		return err

vmdkops-esxsrv/vmdk_ops.py

@@ -1,73 +1,20 @@
 #!/usr/bin/env python
 
 '''
-TOP TODO
-
-0. extract attach/detach code and make sure it work fine without anything else
-1. add WaitForTask in Python  and drop sleep, Also add time track for reconfigure
-2. Add scsi_resync to plugin.go. Check how mount is done in flocker and in VIC
-3. Go over code, and collect TODO in one place (add TODO file)
-4. add simple test script (or sketch it)
-5. Look at cleaning up makefile and adding godeps
-'''
+ESX-side service  handling VMDK create/attach requests from VMCI clients
 
+The requests (create/delete/attach/detach) are JSON formatted.
 
-#
-# ESX-side service  handling VMDK creat/attach requests from VMCI clients
-#
-# The requests (create/delete/attach/detach) are JSON formatted.
-#
-# All operations are using requester VM (docker host) datastore and
-# "Name" in request refers to vmdk basename
-# VMDK name is formed as [vmdatastore] dvol/"Name".vmdk
-#
-# Commands ("cmd" in request):
-#		"create" - create a VMDK in "[vmdatastore] dvol"
-#		"remove" - remove a VMDK. We assume it's not open, and fail if it is
-#		"list"   - [future, need docker support] enumerate related VMDK
-#		"attach" - attach a VMDK to the requesting VM
-#		"detach" - detach a VMDK from the requesting VM (assuming it's unmounted)
-#
-#
-# known issues:
-# - need interrupt handler to restart listen() on control-c
-# - call VirtualDiskManager directly instead of cmd line
-# - todo/tbd below
+All operations are using requester VM (docker host) datastore and
+"Name" in request refers to vmdk basename
+VMDK name is formed as [vmdatastore] dvol/"Name".vmdk
 
-'''
-TODO
-==
-Drop command line and use VDM for disk manipulation:
-spec=vim.FileBackedVirtualDiskSpec(capacityKB=1024, profile = None ,adapterType='lsiLogic', diskType = 'thin')
-vdm = si.content.virtualDiskManager
-vdm.CreateVirtualDisk_Task(name="[datastore] eek/a1.vmdk", datadatacenter=dc, spec = spec)
-vm = findVmByName()... ; vm.config.datastoreUrl[0].name is datastore
-see https://opengrok.eng.vmware.com/source/xref/vmcore-main.perforce.1666/bora/vmkernel/tests/vsan/vsansparse_sanity.py#314
-or even better https://gist.github.com/hartsock/d8b9c56cd7f779c92a78 (fails if exists)
-for examples
-==
-
-make sure backend name is properly calculated -
-OR replace it all with in-guest formatting (but then we'd need to locate the proper blockdevice()
-
-With the current approach, new device is found for mount with blkid -L <volume-name>, which is easier
-
-===
-check size format and handle policy on -o flags
-==
-make sure the NEW disk is really formatted if if it does not have -flat.vmdk
-==
-findChild seems to generate a task - too slow. Find out where and drop
-==
-Pass error as a code in the VMCI package, rather than a hack with server.c:vmci_reply
-
-===
-
- *   TBD: authorization/access config:
- *   - location configuration. And generally, configuration... host profile/adv.conf?
- *   - a local version, talking over Unix Socket, to debug all I can on Linux - with logic and fake responses (part of build unit test)
- *   - an ESX version, talking over http , to debug logic on a random ESX - with full vigor API usage
- *   - an ESX version, with vSocket (C I suppose) connection, to finalize debugging on actual guest/host story
+Commands ("cmd" in request):
+		"create" - create a VMDK in "[vmdatastore] dvol"
+		"remove" - remove a VMDK. We assume it's not open, and fail if it is
+		"list"   - [future, need docker support] enumerate related VMDK
+		"attach" - attach a VMDK to the requesting VM
+		"detach" - detach a VMDK from the requesting VM (assuming it's unmounted)
 
 '''