DeployLinux upgraded all hosts in the production hosting cluster to ESX 3.5 U2 over the weekend.
There was no user-noticeable downtime, or significant issues.
U2 is more of a maintenance release and focuses on bug fixes, but there were some minor feature additions. From our perspective, the most interesting is the ability to fine-tune the HA algorithm to focus more on receiving the heartbeat of individual VM's than hosts....as we've certainly seen cases where an occasional Linux Kernel will lock up. In these cases, we might be able to have VMware automatically restart the VM within minutes rather than waiting for the monitoring system to mark the host down and page our staff who who would investigate the problem and force a manual reboot(e.g. VM restarts in 2 minutes rather than 15?).
We're not 100% convinced that we want VMware automatically restarting VM's as that might slightly impact our ability to track down the cause of a crash, but this may be desired for the most downtime sensitive VM's (e.g. databases). On the other hand, if a database crashed ...there are reasons to want it to restart by hand. There are always trade offs.
In any case, U2 also included several updates to the Virtual Center management application and dozens of bug fixes to ESX itself -- all of which we are very grateful for.
There was no user-noticeable downtime, or significant issues.
U2 is more of a maintenance release and focuses on bug fixes, but there were some minor feature additions. From our perspective, the most interesting is the ability to fine-tune the HA algorithm to focus more on receiving the heartbeat of individual VM's than hosts....as we've certainly seen cases where an occasional Linux Kernel will lock up. In these cases, we might be able to have VMware automatically restart the VM within minutes rather than waiting for the monitoring system to mark the host down and page our staff who who would investigate the problem and force a manual reboot(e.g. VM restarts in 2 minutes rather than 15?).
We're not 100% convinced that we want VMware automatically restarting VM's as that might slightly impact our ability to track down the cause of a crash, but this may be desired for the most downtime sensitive VM's (e.g. databases). On the other hand, if a database crashed ...there are reasons to want it to restart by hand. There are always trade offs.
In any case, U2 also included several updates to the Virtual Center management application and dozens of bug fixes to ESX itself -- all of which we are very grateful for.