The MSA reports the sizes as marketing sizes (Base 10), but there's an option to change it to Base 2, i.e. the real technical size.
Please take a look at e.g. https://support.hpe.com/hpsc/doc/public/display?docId=emr_na-c01762974
That description may not be for for your exact MSA model, but the modification steps should be similar.
André
What you mean marketing size.
MSA shows right one as that is realy size of disk I've inserted there.
3x1.2TB+800GB=4.4TB
When I did created storage on vSphere it showed 4.090 TB.
Where is 300+ GB?
Where is 300+ GB?
It's in the maths, i.e. the conversion from TB to TiB.
Maybe Units converter - convert TB into TiB can help to explain this better.
Simply enter 4.4 TB TiB into the input filed, and you'll see that you end up with ~4TiB.
Btw. it's the same for the disks that you mentioned. A 1.2TB disk has a capacity of ~1.09TiB.
André
So that means vSphere shows in TiB?
Why then not to write TiB intead of TB.
Confusing things.
So that means vSphere shows in TiB?
Yes, they show the real technical size.
Why then not to write TiB intead of TB.
Sorry, I can't answer this.
It's indeed confusing to have such a distinction anyway. In the early IT times 1kB equaled 1,024 Bytes, and nobody ever thought of 1kB = 1,000 Bytes, or s.th. like TB vs. TiB.
André
as Andre has stated this is the difference between a base 2 (binary) Terabyte and a base 10 (Decimal) Terabyte. The former used by computing systems to denote capacity and the later by Marketing departments to inflate their capacity.
Andre, those were purer times before marketing kidnapped, the technical space with verbal athletics.
Running ESXi 6.1 update1 using Pure Storage. I had VVOLs set up and functioning for one small custer in my environment. I migrated a VM to the VVOL datastore and created a couple other VMs experimenting with Storage Policies. Everything seems to be working as expected. The other say when I looked the VVOL datastore now shows 0 bytes available and 0 bytes free. I am unable to browse the datastore, no error..... just never shows any files. I have a VM currently running on the datastore and it seems fine. The storage providers show up and connected. I can see the PE from the hosts. It looks like it is up and connected but somehow the vCenter cannot get the datastore info. I have opened a ticked and sent logs to support twice now. Looking through SPS logs and vvold and nothing jumps out right away.
-Jeff
my ex-colleague has loaded another set of disks (3 disks of raid 5) from another ESXi into the current ESXi before he left.
so i have two datastores, each datastore contains the same system partitions. i assume only one datastore runs the ESXi.
i want to to take out the datastore that do not host ESXi, but i have no idea which device contains the system boot partition.
i could use the command "esxcli system boot device get" to know the uuid of the system boot filesystem.
but how could i base on this uuid to determine which device runs the esxi system and which device could be removed.
thanks,
Check the KB - VMware Knowledge Base. Though it says for Boot from SAN, it is applicable for all types of boot mediums.
Cheers,
Supreet
thank you! that's very helpful!
I ask the question as I did the test below, and found no data block was not freed in lun level after deleting the VM.
if ESX host frees data block in block/LUN level when deleting VMDK file, it can help the storage utilization more efficient.
Did i miss some configuration in vsphere?
1. create a thin-provisioning LUN in my storage array. The allocated size of the new LUN is zero now.
2. mount the LUN in ESXi host as a VMFS data store.
3. create a VM with 40GB eagerly zeroed thick VMDK in the data store.
4. in vsphere UI, the used size of the data store is a little more than 40GB.
5. in the storage array management UI, the allocated size of the LUN is around 40GB.
6. delete the VM.
7. in vsphere UI, the used size of the data store is decreased to less than 1GB.
8. however, in the storage array management UI, the allocated size of the LUN is still around 40GB.
9. I looked into the metadata of the LUN, no data block was not freed in lun level after deleting the VM.
I am doing VASA and VVOL certification for Dell Compellent storage center.
I am facing some issues with the setup and want to have clarity to run our certification test.
Type of Certification:
Setup Details :
Both the ESXi hosts are using FC protocol for Storage array connection.
Software Versions used :
then we have to register VASA Provider with both the VCs. As per our understnading VASA provider can only be registered with one VC. if VASA registration is done on multiple VCs from same Dell Storage manager then there will be issues in creating protocol endpoints and VVOL datastore is not accessible for any of the ESXi host.
Query here is ,Should we be registering same VASA provider with both the VCs or we have to deploy 2 different Dell Storage Managers. and if there is a way to register the same VASA to 2 VCs.
2. If we have 2 different Dell Storage Managers for 2 VCs then while running the Certification test there is a field where we have to mention Dell Storage Manager IP but that is only for one instance and not for two.
How should I setup and meet all the cerfication pre-requisites.
1) Are you referring to automatic unmap with VMFS-6 or are you running manual unmap command?
2) Does your array even support unmap? https://cormachogan.com/2016/12/05/determining-array-supports-automated-unmap-vsphere-6-5/ / Storage reclamation – part 1 – VMware vSphere | Kalle's playground
3) During the unmap, are you seeing an increment in the Delete Stats under VAAIStats in esxtop? VAAI Thin Provisioning Block Reclaim/UNMAP In Action - VMware vSphere Blog
Cheers,
Supreet
Thanks! After running manual unmap command "esxcli storage vmfs unmap -u <UUID>", the storage space was reclaimed.
Does only VMFS-6 support automatic unmap? Can I enable automatic unmap on VMFS-5?
Nope, auto-unmap is available only with VMFS-6. And only if, the storage array supports auto reclaim. Please close the thread if your queries have been answered.
Cheers,
Supreet
1. Do I need to create a VM Kernel in order to achieve this? (like creating an ISCSI VMkernel)
2. Do I have to enable ISCSI in the software HBA if shared storage is NFS?
2. Is NFS recommended as shared VM datastore?
Do I need to create a VM Kernel in order to achieve this? (like creating an ISCSI VMkernel)
Yes, or otherwise have an existing vmkernel interface where this export is accessible.
Do I have to enable ISCSI in the software HBA if shared storage is NFS?
No. iSCSI and NFS are totally different.
Is NFS recommended as shared VM datastore?
It is one possible recommendation, yes. Things always depend on several factors.
I need to implement four HPE ProLiant DL380 machines, two 10 Gb switches for redundancy, and two MSA 2050, where one MSA is offsite for backup purposes - a building across the street. Each host will connect one (or two?) 10 Gb ports of each switch and each switch connected to MSA 2050 redundant controllers using either one or two cables per switch, again for redundancy.
I have been searching for a best practices cabling network diagram but cannot find any that address issue of networking that combine vSphere management tasks, iSCSI, and replication. Can all these things occur via the same network cables?
host 1 -> switch 1
host 1 -> switch 2
host 2 -> switch 1
host 2 -> switch 2
host 3 -> switch 1
host 3 -> switch 2
host 4 -> switch 1
host 4 -> switch 2
switch 1 -> MSA 1 controller 1 port 1
switch 1 -> MSA 1 controller 2 port 1
switch 2 -> MSA 1 controller 1 port 2
switch 2 -> MSA 1 controller 2 port 2
Remote MSA:
switch 1 -> switch 3 -> MSA 2 controller 1 port 1
switch 2 -> switch 4 -> MSA 2 controller 2 port 1
Will this configuration get me vSphere management, iSCSI support, and replication? What improvements could be done?
Thanks,
Kevin
In vsphere 6.5 the maximum size of a vmfs volume is 64TB. I do not see a maximum for a datastore cluster size. Does this mean I could combine 512 (max fc luns per host) 64TB datastores into a datastore cluster? Thanks