vmware: fix volume stats logic

[rohityadavcloud]

During volume stats calculation, if a volume has more than one disk in
the chain-info it is not used to sum the physical and virtual size
in the loop, instead any previous entry was overwritten by the last disk.

Fixes #3416

Types of changes

• Breaking change (fix or feature that would cause existing functionality to change)
• New feature (non-breaking change which adds functionality)
• Bug fix (non-breaking change which fixes an issue)
• Enhancement (improves an existing feature and functionality)
• Cleanup (Code refactoring and cleanup, that may add test cases)

[rohityadavcloud]

@blueorangutan package

[blueorangutan]

@rhtyd a Jenkins job has been kicked to build packages. I'll keep you posted as I make progress.

[blueorangutan]

Packaging result: ✖centos6 ✖centos7 ✔debian. JID-91

[rohityadavcloud]

@blueorangutan package

[blueorangutan]

@rhtyd a Jenkins job has been kicked to build packages. I'll keep you posted as I make progress.

[blueorangutan]

Packaging result: ✔centos6 ✔centos7 ✔debian. JID-95

[rohityadavcloud]

@blueorangutan test centos7 vmware-65u2

[blueorangutan]

@rhtyd a Trillian-Jenkins test job (centos7 mgmt + vmware-65u2) has been kicked to run smoke tests

[anuragaw]

LGTM. Haven't tested but code looks solid.
Would be great, @andrijapanic if you can quickly test and LGTM this.

[andrijapanicsb]

Can't comment atm, but @rhtyd this was an issue with both full clones and linked clones.

I'll be able to test this after 17th...

[blueorangutan]

Trillian test result (tid-107)
Environment: vmware-65u2 (x2), Advanced Networking with Mgmt server 7
Total time taken: 33310 seconds
Marvin logs: https://github.com/blueorangutan/acs-prs/releases/download/trillian/pr3473-t107-vmware-65u2.zip
Smoke tests completed. 66 look OK, 6 have error(s)
Only failed tests results shown below:

[shwstppr]

@blueorangutan package

[blueorangutan]

@shwstppr a Jenkins job has been kicked to build packages. I'll keep you posted as I make progress.

[blueorangutan]

Packaging result: ✔centos6 ✔centos7 ✔debian. JID-97

[rohityadavcloud]

@blueorangutan test centos7 vmware-65u2

[blueorangutan]

@rhtyd a Trillian-Jenkins test job (centos7 mgmt + vmware-65u2) has been kicked to run smoke tests

[shwstppr]

@rhtyd Still not seeing correct results. With VMware 65u2, I don't think code is taking care of -delta.vmdk. Please see screenshots,
From NFS primary storage. -delta.vmdk file size increases as I write data in instance.

Instance storage is almost full

But UI and API does not show correct physical size

[blueorangutan]

Trillian test result (tid-113)
Environment: vmware-65u2 (x2), Advanced Networking with Mgmt server 7
Total time taken: 33304 seconds
Marvin logs: https://github.com/blueorangutan/acs-prs/releases/download/trillian/pr3473-t113-vmware-65u2.zip
Smoke tests completed. 68 look OK, 4 have error(s)
Only failed tests results shown below:

[rohityadavcloud]

@blueorangutan package

[blueorangutan]

@rhtyd a Jenkins job has been kicked to build packages. I'll keep you posted as I make progress.

[blueorangutan]

Packaging result: ✔centos6 ✔centos7 ✔debian. JID-102

[rohityadavcloud]

@anuragaw @shwstppr anyone wants to take this one?

[shwstppr]

@rhtyd will be looking into this

[shwstppr]

As reported in #3473 (comment), ROOT-*-delta.vmdk file size increases whenever something is written to instance disk. But this vmdk file is not taken into account while calculating volumes' physical size.
I get following volume info for one such volume,
{"diskDeviceBusName":"ide0:1","diskChain":["[729e08870a5d311c92d455db46da45d1] i-2-3-VM/ROOT-3.vmdk","[729e08870a5d311c92d455db46da45d1] 3ca2d374b8083bfda1591a3e3757db70/3ca2d374b8083bfda1591a3e3757db70.vmdk"]}
Dring my debugging I couldn't find any ACS code that deals with ascertaining this disk chain and it appears it is done by VMware library class VirtualMachineDiskInfoBuilder.
Any help here @nvazquez @rhtyd

[shwstppr]

From vCenter,

From storage,

I've verified with debugging https://github.com/apache/cloudstack/blob/master/vmware-base/src/main/java/com/cloud/hypervisor/vmware/mo/VirtualMachineMO.java#L2597-L2609, VirtualDisk points to ROOT-5.vmdk and not to ROOT-5-delta.vmdk. Therefore only two vmdk files in the chain,
{"diskDeviceBusName":"ide0:1","diskChain":["[b5fe57d739a33733bdd825fa62e86d4d] i-2-5-VM/ROOT-5.vmdk","[b5fe57d739a33733bdd825fa62e86d4d] 9b41f3d52a7f36678e302280f7ca2bbb/9b41f3d52a7f36678e302280f7ca2bbb.vmdk"]}

Even with ROOT-5.vmdk file size is different from vCenter and storage. Do we need to explore a different method to get volume usage on VMware?

[rohityadavcloud]

Merging this based on lgtms and reviews. It's possible that more fixes need to be done, but at least this change ensures that multi-chained disks are not ignored. I'll keep the original issue open but close/merge this PR.

[DaanHoogland]

logic looks sane, i hope we can rely on the vmware logic to be as sane (i.e. chainInfo works as intuitively and consistently as it seems from this code)