So I had 4 VMs running just fine on my Windows 2003 server (VMware Server host). I decided to convert a single growing VMDK (disk type 0) to a pre-allocated VMDK (disk type 2) to prevent fragmentation of the VMDK on the host disk (yes, run defrag after I pre-allocate). The moment I executed the vmware-vdiskmanager command, all the VMs halted.
I couldn't figure out what happened the first time I did this. Nothing failed other than the VMs stopping in their tracks (no disk, network, cpu activity).
Subsequent times running vmware-vdiskmanager (because I have a couple single growing VMDKs that I'm converting), I noticed that upon execution, all memory (in RAM) consumed by any vmware-vmx.exe process (the actual VM process) is flushed down to (or near) 300-400KB. Virtual memory consumption remains unchanged. Note, this is all per Windows Task Manager. I found that after I execute it, if I change the priority on the vmware-vdiskmanage.exe process to Below Normal, my VMs come back to life faster, but they are still halted for at least 5 minutes. The VMs, however, don't seem to notice that they stopped. Everything still proceeds to run seamingly normally after they resume.
My question is WHY? Am I doing something wrong? What is the right way?
Host stats
HP ProLiant ML370 G4
4GB RAM
Intel Xeon CPU 3.40GHz (2 CPUs, 4 logical CPUs with H/T)
3 NICs
Windows Server 2003 Standard R2
VMware Server 2.0 build 122956
Storage:
Internal, Smart Array 6400
OS - 34GB, RAID 1+0, 2x 36GB 15K
Datastore - 684GB, RAID 5, 6x 146GB 10K
