ESXI Ultra-Annoying Extra-Low Datastores Transfer Rates workaround!
Just when I’m finally ditching ESXi because I absolutely hate some of its stupid limitations, I accidentally stumble on a workaround for one of its most annoying issues. Anyone running a home lab or production ESXi with tons of data in the datastore knows how unbearable it is waiting to get data out. Transfer speeds are painful, to put it mildly. Anyway, I’m migrating my server to Proxmox and moving the VMs over. So I’m backing up a few TBs of images to an ext4 drive instead of vmfs6. I started this on a Tuesday, hoping to be ready to install everything by Saturday, given the famously endless low transfer speed.
So, I plugged an SSD into the ESXi box, but then remembered it doesn’t read ext4. I logged into one of the VMs, set up a passthrough vmkfstools -z /vmfs/device/device_id /vmfs/volumes/datastore/server/disk.vmdk mounted it in Ubuntu, partitioned it, formatted it, and mounted it to a folder. Next, I mounted the ESXi datastore via sshfs sshfs [email protected]:/vmfs/volumes/datastore1 /mnt/datastore1 and fired off an rsync between them as a background service, expecting it to run for days. I stepped out of the office, cracked a joke with my kid playing Fortnite, let the dog out, grabbed a glass of water, and when I got back, I noticed something weird. It was already on the 3rd server with about 300GB copied. I checked the transfer rate, and to my surprise: 100+MB/s!
Just a heads-up! Hope it’s useful!
EDIT: Ah, a poor man’s joy is always short-lived! Not even an hour into the transfer, the speed tanked back down to the usual 2 to 6MB/s. I have a conspiracy theory that this is on purpose, because come on, there’s just no way! Ahahahaha!