Full datastore when transfering data

Hi all,

I'm having a serious problem with datastore capacity when I transfer a huge amount of data to a virtual machine by SSH. The datastore has 511GB of total capacity. During the transfer proccess the virtual machine stops, just when the DataStore gets full. So the only solution is removing other virtual machines from the DataStore, but it's not enough.

We are using VMware ESX 3.5 and the virtual machine is Ubuntu Server. Here the configuration of the VM:

Memory: 1024MB

CPUs: 1

HardDisk1: 6GB

HardDisk2: 52GB

HardDisk3: 230GB

HardDisk4: 256GB

Storage distribution and capacity in VM View:

# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/VG00-VG00_ROOT
                      3.4G  2.6G  621M  81% /
none                  495M  220K  495M   1% /dev
none                  500M     0  500M   0% /dev/shm
none                  500M  524K  500M   1% /var/run
none                  500M     0  500M   0% /var/lock
none                  500M     0  500M   0% /lib/init/rw
/dev/mapper/VG01-VG01_home
                      5.4G  276M  4.8G   6% /home
/dev/mapper/VG00-VG00_TMP
                     1008M   34M  924M   4% /tmp
/dev/mapper/VG00-VG00_LOG
                     1008M   53M  904M   6% /var/log
/dev/sda1             122M  118M     0 100% /boot
/dev/mapper/VG02-VG02_home_dpto
                      479G  219G  236G  49% /home-dpto

HardDisk3 and HardDisk4 are mapped into a LVM volume (VG02-VG02_home_dpto) and its free space is 236GB. DataStorage free space: 4,56GB. I don't understand why. The problem may be caused by automatic generating new VMDKs:

*ficheros00.platform.core.vmdk ->6GB

*ficheros00.platform.core_1.vmdk ->54GB

*ficheros00.platform.core_2.vmdk ->241GB

*ficheros00.platform.core.-000002.vmdk ->24GB

*ficheros00.platform.core_2-000001.vmdk ->166GB

At the first time, I got 20G of free space of the DataStore deleting some residual VMs. Then I tried to copy a 15GB folder through the network by SSH from another file server. This was the occupancy at the first time: (just compare)

*ficheros00.platform.core.vmdk ->6GB

*ficheros00.platform.core_1.vmdk ->54GB

*ficheros00.platform.core_2.vmdk ->241GB

*ficheros00.platform.core.-000002.vmdk ->10GB

*ficheros00.platform.core_2-000001.vmdk ->164GB

It seems 000002.vmdk and 00002.vmdk are growing during the file transfer but they don't consolidate data. I don't know whether this is a normal behavior. I think the space used for  ficheros00.platform.core.-000002.vmdk ->24GB and ficheros00.platform.core_2-000001.vmdk ->166GB should be included within the ficheros00.platform.core_2.vmdk ->241GB and at the end of transfering data they shoud dissapear themselves.

We calculate the initial free space before allocate hard drives to the virtual machine.

Currently, in this dataStore we have 2 VM:

[root@hp585-01 eva4400:-02]# du -sh ficheros00.platform.core./
472G    ficheros00.platform.core

[root@hp585-01 eva4400:-02]# du -sh lucus01.labo.core_1/
36G     lucus01.labo.core_1

Any help would be appreciated. Thank you very much.

Best regards,

           Harry.