Virtual Volumes FAQ
Please keep in mind that the answers provided in these sections are not comprehensive or detailed. Rather they are intended to provide quick reference to some common questions. Please see the Implementation Guides for more detailed explanations and guides for vVols.
What Purity version is required for vVols?
Purity 5.0.0 and later is required. 5.3.15+ is the minimum recommend version and Purity 6.1.10+ is the recommend version.
What FlashArray models are supported with vVols?
400 series, //M, //X, //XL and //C
What version of ESXi and vCenter is required?
ESXi 6.5 Update 1 and later and vCenter 6.5 Update 1 and later.
vSphere 6.7 U3 or higher is the recommended release.
See the VASA and vVols Related Fixes by ESXi Release KB for more detailed information about specific fixes or features for vSphere releases.
Are there additional pre-requisites?
Configure NTP for ESXi, vCenter and the FlashArray. Such that there is no time skew between any of the three.
Port 8084 need to be open for vCenter, ESXi and the FlashArray for the VASA Service to be accessible.
Is FlashBlade Supported?
No. There are no plans to support vVols on FlashBlade at this time.
Can a FlashArray be registered with more than one vCenter?
With Purity 5.3.0, certificates can be imported to VASA, which means that multiple vCenters that are not in ELM (Enhanced Linked Mode) can register with a single FlashArray. See this KB for more information.
If the FlashArray is under Purity 5.3.0, then the FlashArray can only be registered with multiple vCenters if they are in Enhanced Linked Mode (ELM) and share the same SSO. If the vCenters are not in ELM and have their own SSO, then registering another vCenter will invalidate the registration in the existing vCenter.
What is VASA?
VASA stands for the VMware vSphere API for Storage Awareness. This is a service that is provided by the storage platform for communication from VMware to the array. The FlashArray VASA provider converts VMware-issued API calls to FlashArray-specific requests.
What version of VASA do we support?
VASA version 3.0 in Purity 5.0.0 to 6.0.x. This provides replication support for vVols.
Beginning with Purity 6.1.3 and higher the FlashArray VASA Provider also supports VASA version 3.5.
How is VASA installed/configured?
Installing Purity 5.0 or higher will automatically install and configure VASA on both of the FA controllers. No configuration is required.
How do you register VASA with vCenter?
You can use the FlashArray vSphere Plugin, use the vSphere Client itself, PowerCLI or vRO.
What is the URL for the VASA providers?
You must use this form for the URL:
https://<IP of CT0>:8084
https://<IP of CT1>:8084
Do NOT use the VIP.
Can I use the VASA provider with a different port?
No, it requires 8084.
My registration is failing. Why?
DO NOT use the virtual IP or a FQDN. It must be the controller eth0 IP. Should Purity 5.3.0+ be in use, the customer can import their own certificates and use eth0 or eth1.
Make sure vCenter has access over IP to the FlashArray management ports on the controllers. TCP 8084
Should I register both VASA providers?
Yes always. This supplies high availability of the providers in case of a controller loss or reboot.
Are the VASA providers active/active?
Yes, but vCenter will only use one at a time—the one it uses is the first one it sees for a FA.
How does a customer restart a VASA provider?
They cannot (outside of power-cycling a controller).
Does VASA run in Purity//Run?
No, it is a service running in the controllers, like the GUI, Purity etc.
What is a protocol endpoint?
This is the physical connectivity to the array. Before a vVol datastore can be used by a host you must connect a PE to it.
How do I create a PE?
Generally, there is no need to. Upon the first time VASA is registered with vCenter the VASA provider creates the volume called pure-protocol-endpoint.
Do I need more than one PE?
No. One PE per FlashArray is fine. More can be created, though through FA CLI only:
purevol create --protocol-endpoint protocol-endpont-name-01
How do I connect a PE?
You must use the CLI and connect it to a host group or host, GUI support for connecting and disconnecting PEs was added in Purity 5.3. You can use the vSphere Web Client Plugin 3.0. The Create Datastore workflow allows mounting of a storage container and will automatically present the PE to the host or cluster.
My Cisco UCS ESXi host does not see the PE I presented. Why?
Most common reason is that the fnic driver is out of date. https://kb.vmware.com/kb/2111492 Follow this:
/var/log/vmkernel.log file on the ESXi host will show the following
vVol PE warnings when the “Rescan Storage” is initiated:
2018-01-09T18:04:42.098Z cpu5:65799)WARNING: ScsiPath: 705: Sanity check failed for path vmhba0:C0:T1:L1.
My ESXi host does not see the PE I presented. Why?
If it not UCS, the most common reason is that they forgot to provision the PE to the host or host group, or the zoning is not complete, or iSCSI is not configured on the host.
What is a vVol datastore?
This is best answered by what it is NOT. It is not a LUN, it is not a file system, it is not a volume.
It is a capacity quota from a given FlashArray. There is no direct relationship to using specific FlashArray features or performance to a given vVol datastore.
How big is a vVol datastore?
From Purity 5.0.0 to Purity 6.4.0, the vVol datastore defaults to 8 Petabytes. With Purity 6.4.1 and later, the default is 1 Petabyte because of this issue.
How many vVol datastores are supported?
From Purity 5.0.0 to Purity 6.4.0, 1 per FlashArray. In Purity 6.4.1 and later, there can be 1 vVol datastore associated with every pod on the FlashArray.
Can you resize a vVol datastore?
How do you resize a vVol datastore?
In order for a customer to resize a vVol Datastore, they will need to open a Support Case with Pure Storage and request them to resize the VASA Storage container.
What is the difference between a storage container and a vVol datastore?
Nothing really. A storage container gets mounted in ESXi as a vVol datastore. So a vVol datastore is really just a name for the mount point of a storage container.
I can mount a vVol datastore but it is inaccessible to the host or showing zero capacity. Why?
The most common reason is the PE is not presented.
My PE is presented to the host, but the vVol datastore is still inaccessible. Why?
Sometimes you need to just simply restart the vvold on the ESXi hosts. SSH into the host and run:
If that does not work, it could be a cert issue, run:
Sometimes you may need to refresh the Storage Provider Cert in vCenter or more often is the case, force vCenter to refresh the root CA and CRLs to the ESXi hosts. See this section of a KB for more information.
What gets created when I create a VM?
Every VM will get a volume group. For each VM there will be a config vVol and a data vVol for every virtual disk. When powered on, a swap vVol will be created. Potentially there will be memory vVols too.
What happens if I manually delete the volume group?
Nothing really. Though it is used to represent the VM object, so it will be a bit harder to manage directly from the array. We recommend not deleting it.
What happens if I manually rename the volume group?
Then the volume group is renamed--nothing else is affected. If you renamed the VM, it is a good idea to rename the volume group--this does not happen automatically.
What happens to the volumes and volume groups if I rename the VM in vSphere?
Unfortunately, the names will be mismatched, as VMware does not notify us of rename operations. This is something Pure is working on to see if we can do something to improve it. Currently, you would need to rename the volume group to new name manually.
How are vVols identified?
Internally, Pure Storage uses tags. Pre Purity//FA 6.0 Tags were not exposed to the customer. In Purity 6.0 and later, VASA tags are behind an internal namespace used for VASA. While customers can create their own namespaces or use the default namespace to tag volumes and not impact the tags used by VASA.
Can I convert a regular volume to a vVol?
There is no real “conversion” but it does need to be assigned a vVol ID, which requires support intervention today. The easiest option is to create a new vVol through VMware and purevol copy the old volume to the new vVol volume. Make sure they have identical capacity.
If I delete a VM how do I recover it?
There is a detailed KB that works through general vVol recovery workflows. See that KB for more detailed information.
If I delete a data vVol how do I recover it?
There is a detailed KB that works through general vVol recovery workflows. See that KB for more detailed information.
Do I have to use SPBM (Storage Policy Based Management) with vVols?
No. If you do not plan to use replication or snapshot policies, or need compliance checking, you do not have to assign policies to VMs.
At what granularity can I assign policies?
You can assign policies at the VM level, or individual virtual disk level.
If a VM is out of compliance how do I fix it?
You can manually reconfigure the vVols on the array, or the better option is to re-run the compliance wizard and have VMware reconfigure them. If the current FlashArray cannot satisfy the policy, you will need to Storage vMotion the VM to an array that can.
How do I fail over virtual volume VMs?
This can be done with vRO, PowerCLI, or any VMware tools that can use the SDK. Site Recovery Manager 8.3 added support for vVols array based replication as well. Pure does require using Purity 5.3.6+ to use SRM with vVols.
Should I replicate vVols manually or assign replication via storage polices?
Always use VMware storage policies to assign replication to vVol-based VMs. This will have VMware tell the FlashArray where to put the vVols (what protection group) and more importantly, VMware now knows they are replicated and can fail them over.
Please see our vVols Replication Deep Dive KB for more detailed information about vVols with replication.
Is Site Recovery Manager supported with vVols?
Yes, with the FlashArray VASA Provider 1.1.0 (Purity 5.3.6+), Pure Storage supports vVols and SRM 8.3. => https://blogs.vmware.com/virtualblocks/2020/04/14/announcing-vmware-srm-integration-pure-storage-arrays-vsphere-virtual-vvols/
Does vRealize Operations Manager support vVols?
VMware added support for vVols in vROps beginning with vROps version 8.1.
Pure Storage added support for vVols in the management pack with version 3.1.0. However, Pure recommends to run the 3.2.0 management pack or higher.
SRM device discovery fails after upgrade to Purity 5.0
Due to the larger LUN IDs with vVols, the older SRA fails discovery when seen, even if you only want to use SRM with VMFS. You must update to our latest SRA 2.0.8. This can be downloaded from VMware’s site.
vRealize Operations FlashArray Management Pack fails after Purity 5.0 upgrade
Due to the larger LUN IDs with vVols, the older mgmt. pack chokes on this when seen. You must update to our latest pack 1.0.152. This can be downloaded from VMware’s Solution Exchange site.
Does vVols work with ESXi 6.7 and vCenter 6.7?
VASA 1.1.0 is on the VMware compatibility guide for 6.7 and 7.0. VASA 1.0.2 is listed as supported with vSphere 6.5 and 6.7.
What version of the vSphere Plugin is needed for vVol support?
3.0. This is supplied with Purity 5.0 and later.
Does the vRealize Orchestrator plugin work with vVols?
Yes. There is a vVols workflow package available as well.
Is there any integration in Pure1 with vVols?
Yes, refer to the Pure1 KBs for more information.
PowerCLI cmdlets are failing, why?
You must be using PowerCLI 6.5.4 or later.
Does MSCS (Microsoft Cluster Service) / Windows Server Failover Clustering (WSFC) work with vVols?
Yes, the FlashArray supports MSCS (2008 and earlier) or WSFC (Windows 2012+). Beginning in ESXi 6.7 VMware started to support MSCS with vVols (allowing SCSI-3 RESERVATION requests to subsidiary LUNs). While support was added in ESXi 6.7 GA, an issue was identified early on within VMware engineering that prevented Microsoft Clustering Services from working as expected with vVols. A fix for this issue has been provided in ESXi 6.7 U2 and is required for MSCS / WSFC functionality with vVols.
In order to ensure a smooth integration Purity 5.1.11+ or 5.3.6+ are recommended, though not required.
Purity Feature Interoperability
Can vVols run on a FlashArray with ActiveCluster?
Yes, but with a caveat. vVols can run on FlashArrays enabled for ActiveCluster, however any protocol endpoints, volume groups or vgroup volumes can not be added to a Pod.
Can vVols be stretched or protected by ActiveCluster?
No, as stated above, volume groups, volumes in volume groups and protocol endpoints can not be placed in a FlashArray pod. Which means that these objects can not be stretched. This in turn means that vVols can not be stretched and used within an ActiveCluster pod.
Can vVols run on a FlashArray with Purity//Run Enabled?
Yes, vVols can run on an Array with Purity//Run Enabled on starting on 5.1.1 and above.
Does VM Analytics Gather vVols related Statistics?
Yes, see the Pure1 KBs for more information.
Can an Array have multiple vVol Datastores?
No, at this time Pure supports having only one vVol Datastore per Array.
Can an Array be connected for vVols with Multiple non-linked vCenters?
Yes, see the section about this above.
What are the current Array Limits?
Object limits depend on Purity Version and Hardware Model. See the object limits KBs for the specific limits of the that combination.
- Large FlashArray Model Limits
- Medium FlashArray Model Limits
- Small FlashArray Model Limits
- FlashArray//XL Model Limits
- FlashArray//C Model Limits
Reminder that a volume = a vVol. A volume group = VM.
These are some of the features and integrations that Pure Storage is actively working on for vVols.
- Decreasing the stun time during managed snapshots.
- Increasing scale for vVols objects on FlashArray devices.
- New VM undelete and revert workflows for the vSphere Plugin.
- ActiveCluster Support for vVols.
- Multiple storage container support.
- And more that will be shared in the future.