Check (warn) if instance has a pending reboot due to changed configuration

Currently, if modifying the configuration of an instance via gnt-modify ... the user is told that, for the change to have any effect, the instance needs to be rebooted. Probably, we should introduce a check for this in gnt-cluster verify, to be able to find instances with pending reboots due to changed configurations.

Probably, this is as easy as comparing mtime and start_time:

if
   mtime > start_time
then
   warn
endif

[zen2]

I think that will be a real good feature because here for our usage it’s something really needed. Sometimes we rebalanced CPU/RAM for lot of VM but we will reboot them later to don’t impact their usage. And very often we forget one or two VMs to reboot because there is no centralized way to know it. I’ve even a script based on gnt-list that do a diff about current running parameters against definition parameters to check if some VM need to be rebooted. So to have this feature natively would be great.

[iustin]

Thinking some more, I think restart with custom parameters should have it’s own warning, and should not affect timestamp change (related to modification).

So the semantics I’m proposing:

• if not started (standard) since last modify, warn
• if started with custom parameters (unrelated to modifications), show a warning or info

Thoughts?