TimothyBates wrote:
I know VMware is responsive to such concerns, because one of the things that did come through the last View release that we pushed for was the "Minimum number of ready desktops during Composer Operations" so that recomposes wouldn't just chew through all available VMs in a desktop pool before readying other replacement desktops.
Unfortunately, this feature does not completely work. I noticed during a recompose of a pool yesterday that as soon as a VM gets to the final "powering up" stage during recomposing, View will assume that the linked clone is now available and then set more VMs into maintenance mode and recompose them. Depending on how long the final power-up and agent check-in takes, it could mean a pool is unavailable for use for a period of time on small pools. Worse, if the VM fails for whatever reason (say your final master prep step was running the wrong script that disabled the firewall and you use HTML Blast) and the task was set to continue on error your entire pool could be unusable due to "Protocol Failure."
Anyone else notice this behavior?