There is no more space for virtual disk ServerName_2.vmdk. You might be able to continue this session by freeing disk space on the relevant volume, and clicking Retry. Click Cancel to terminate this session. Time: 30/05/2014 1:16:20 AM

Question

Recently, our mail server crashed at about 7pm one night, with the error 'There is no more space for virtual disk ServerName_2.vmdk. You might be able to continue this session by freeing disk space on the relevant volume, and clicking Retry. Click Cancel to terminate this session.'
When we click Retry, the server starts up OK.
There are no snapshots listed in Snapshot manager for any of the virtual machines on the host.
There is also free disk space available on the host and for the VM with the disk errors.
This happened at least three more times, often at bad times. Each time, we were able to click 'Retry' and the disk/system would allow the VM to start-up successfully.
I checked the Forums, the VMware support articles and the internet as I had not seen this problem before. I have completed the VSphere 5.1 - Fast Track course and this issue was NOT covered in the training.
Most of the advice on-line and even that on the VMware web-site was pointing to snapshots being the cause of this issue. There are no snapshots enabled and I cannot see evidence of snapshots ever being used.
- We are running VMware vsphere (5.1.0) and there are (were) 4 virtual machines running on the ESX host. We are using the free version of VMware/ESXi.
- The Hard disk types we are using for this Virtual Machine are 'Thin Provisioned'.
- There are 4 [Thin Provisioned] Hard Disks for this virtual machine.
- There are 6 CPUs
- There is 20GB of RAM (memory)
- The VM is running Windows Server 2008 R2 as the guest/VM operating system. It is an Exchange 2010 SP1 mail server. There is plenty of available disk space on all the drives. The [Exchange] log files are cleaned out regularly (automated).
I decided to move one of our non-critical servers off this host and on to another host to see if this helped the problem. This took quite some time, as we are not using HA or vmotion, nor do we have VCenter Server...nonetheless, I finally managed to get the non-critical server on to another host (n.b. This was a much smaller machine with less virtual resources assigned to it).
After moving the non-critical server off this host, we decided to monitor the Host and see if the issue resolved itself.
I checked the host about 6-10 times a day, from first thing in the morning till last thing at night - monitoring the performance of not only the Virtual Machine, but the ESXi host also.
There were no adverse performance issues. The only thing I did note, was in the Summary page on the ESX Host, under Storage, was If I right-clicked on the datastore and clicked refresh, then the free disk space would drop (ie from 140GB to 125GB).
After monitoring the host and VM for about 2 weeks, we did NOT have another instance of the above error.
Sorry for the long winded post, but I wanted to give as much detail given this error has been raised before and snapshots are usually blamed as the cause.
My question is this:
If the ESX host had plenty of available disk capacity and there were no snapshots enabled on the VM (or any other VM's on the same host), then why did our Virtual machine crash with the error that 'there is not more space for virtual disk Servername_2.vmdk'?
How do we prevent this issue from happening if we don't know the underlying cause?
I would greatly appreciate any advice or suggestions.
If I have not provided enough info on the specs or environment, please let me know and I will provide more information.
Thanks all,
Kurt

Answer 1

The type of storage is really based on your requirements, and your ability to withstand downtime.
iSCSI as you are using with a NAS such as Synology or QNAP.  NAS Selector - Support - Synology - Network Attached Storage (NAS)
I wouldn't use iSCSI for Exchange or any database.  It's a bit slow.
Do you have a single physical host?  Then I'd probably to an external direct attached storage.  This would be a card inserted in your host server that gives you multilane SAS/SATA connectivity (www.techcable.com/SAS-SATA/SAS-SATA.pps) and an external disk enclosure/array.
For multiple hosts to a single array, I recommend a fibre channel connection to a FC capable switch, and on to a FC connected array.
We used to use a HP P2000 (on old G1), but it's since been retired.  Worked pretty well once firmware was upgraded.  http://www8.hp.com/us/en/products/disk-storage/product-detail.html?oid=4118559#!tab=features.  They can be connected via iSCSI, Fibre Channel or 6GB SAS so they are flexible and reasonably priced.
Recommendations:
     Use RAID 6 with your large disk arrays.  With large disks there is a measurable failure rate when rebuilding a failed RAID5 array based on MTBF.
     Use smaller 15K disks in RAID 0+1 for speed on databases/Exchange.
     Use slower 7.2K disks in RAID6 for file storage.
We are a small hospital and we have 3 VMware servers with dual CNA (FC and Ethernet in a single twinax cable) connections to 2 redundant Cisco Nexus 5K switches and then 4 Fibre Channel connections to an EMC VNX 5300.  It's extremely fast with about 50 virtual servers, but was quite an investment.  One thing we don't have to worry about is down time.  If there ever is an equipment failure, we have redundant everything, including power split between two UPSs.
Our VNX has 3 tiers of performance.  3 100GB SSD "Fast Cache" in RAID 1 with hot spare, to keep the most used data ready, but it's not really a tier, however one could be built utilizing the same disks.  A second tier is performance tier with a 8 600GB RAID 0+1 and hot spare.  The third is a bunch of 7.2K 3TB disks in RAID6.  The VNX autotiers, placing data on disks depending on where it's needed.  The volumes are sliced and diced automatically in the background to make this happen and we never have to touch it.  I used a demo of Solarwinds Storage Manager to monitor performance for a while and the utilization was always low, meaning all data access was fast, througout the day.
D