i am a new esx user and i am having a problem with space on my datastore, it seems to be running out very quickly and i cant seem to find the reason. I am running 7 vms its a test enviroment with several exchange servers and as i populate them the space continues to disappear. I disabled all the logging options i could find but the space just keeps running out i had some snapshots that i deleted to make room but the space has since disappeared. by my calculations i should only be using 192gb of the 288gb available. so my question is where is the space going? from my understanding space for the vms is reserved when u create them or am i wrong?
Yes all space is allocated for your vm hard drives when they are created. One thing that can take alot of extra space on your vmfs volume is vswp files. These files are created if you do not have memory reservations. So if you have a VM with 2GB or RAM, when it is powered on it will create a 2GB vswp file in the vm's directory on your vmfs volume. So if you edit your VM, then go to Resources tab and set a memory reservation for your vm it will not create the vswp file. IF you create a 2GB reservation it will create no vswp file, if you do not want to create that large of a reservation you can go smaller, for example a 1GB reservation will only result in a 1GB vswp file. Vswp files can add up if you have alot of vm's that have alot of memory assigned to them and they eat up valuable space on the vmfs volume. If you have 7 VM's that have 2GB of memory each with no reservations that will add up to 14GB of disk used by vswp files.
Also you can browse your datastore to look at what is taking up space. To do this click on your ESX server in the VI client. Then on the summary tab right click on your datastore and select Browse Datastore. You can then browse the directories and see all the files and their sizes. Make sure you expand the size column so you can see all the numbers. Make sure there are no extra vmdk files that shouldn't be there.
Also when you browse the data store you will see two vmdk files for each vm hard drive. They will show as the same size when you browse the the VIC. In reality the -flat file is the size of your VM hard drive. The non -flat file is small and is the descriptor file for vmdk file. I'm not sure why it shows as the same size in the VIC but it is not.
You can also browse your whole ESX server file system including vmfs volumes using WinSCP which uses a Windows explorer type interface. This will show you the true size of the non -flat vmdk files
http://winscp.net/eng/index.php
Also if you log into the ESX server console you can use df -h to view partition space and vdf -h to see vmfs volume space
If you find this post helpful please award points...thanks
i did actually manually go through the datastore and check out the size of the files which is how i got my totals as to what my storage usage should be but it doesnt add up to what it actually is. So i dont know if theres something else that i should be taking into account that uses more space and builds as the vm gets used or what. I did not know about managing the vswp files though, thanks.
How much space are you off by? Try doing vdf -h or using Winscp to browse the volume also, the VIC is not always accurate.
80gb give or take as long as i didnt miss count it seems like a big number i know that space is missing though because ive deleted at least 4 or more snapshots a day or so ago and that released to me at least 30-40gb free space and now im down to 14 or so without adding any new vms but with just populating my current vms with data
Are you sure there are no leftover snapshot entries for any VM's that did commit successfully? The space has to be there somewhere it's just a matter of finding out what is using it. You have checked all the VM directories for any extra files? Are you deleting the snapshots by using the Snapshot Manager?
there are some snapshots left but i had taken multiple snapshots and i deleted one layer with the snapshot manager and did get that space recovered but something is continually eating at any space that i manage to free up
in fact i just checked im down to 9gb free space
Message was edited by:
mmbigwolf
If the block size on the volume is set to 8mb, and you have many smaller files, you could be loosing space on account of that..but 80gb seems alot for that theory
its actually set at 1mb
So every VM has log files, swap space minus whatever is reserved for it, snapshots, VMDK files, a few headers and other such info files like the VMX. Have you tried listing all of the files on your VMFS volume and export that to text, then check it again in a few hours. Whichever of those files change is probably your culprit, since the VMDK and VMX files should not really change in size.
Kix
Did you ever find a solution to this? I'm have the same issue. Thanks.
What version of ESX are you using?
I had a problem because I used to mount a remote diskstore under /mnt and I exported my disk files there. Sometimes the connection would drop and I stored the diskfiles directly on / filesystem \!!!
