Description of problem: The "Virtual Machines" counter on the top of the dashboard shows a yellow "warning" triangle when it detects VMs that are in the process of starting up or shutting down. This as probably not something an admin looking at the dashboard should be warned about. Version-Release number of selected component (if applicable): 1.0.0-20160705git3e1e99 How reproducible: Might be difficult to time things right Steps to Reproduce: 1. Have a system where VMs constantly go up and down 2. Look at the dashborad, it will eventually detect VMs that are not 'up' or 'down' and will show then as warning.
For VM objects, the current mapping is following [1]: UP = [Up, SavingState, RestoringState] WARNING = [PoweringUp, Paused, MigratingFrom, MigratingTo, Unknown, WaitForLaunch, Suspended, PoweringDown] DOWN = [Down, NotResponding, RebootInProgress, ImageIllegal, ImageLocked] [1] org.ovirt.engine.ui.frontend.server.dashboard.maps.VmStatusMap So if `PoweringUp` or `PoweringDown` shouldn't be in WARNING category, where else should it be?
Well, powering up might fit up better than warning, and powering down to down. Although I don't have a strong opinion on that one. Moran?
IMO warning should be reserved for stuff we expect the admin to take action to resolve - so VMs should probably show up there for the following statuses: - Paused - but perhaps only if it was paused for storage failure and not because of the user requesting to start in paused mode - Unknown - Down - if went down because of some system failure - NotResponding - ImageIllegal - not sure how you get this one, but sounds bad enough to alert the admin
In general, Dashboard UI just reflects the status categories computed on Engine. Currently, VM inventory card shows 3 status categories: up, down, warning. It's not a tech problem to add more status categories or reorganize existing ones. We'd need a specification for this kind of change, though.
I don't think more status categories are needed. From my personal experience (which is gained from maintaining a couple of moderately large RHEV instances, but not with actual customer data, which I'd love to see PMs contribute), "up", "down" and "warning" is sufficient. Maybe I was not clear in Comment #3. The list I specified there was the list of statuses in engine that I think should be mapped to "warning". I talked briefly with the virt team before posting, the statuses I listed should not be hard to gather from data given by the engine.
Let's wait for PM feedback.
Generally speaking, i would reffer the suggestion in Comment #3 as error state and not warning. I think there is a lot of logic in this suggestion. but AFAIR the decision around states translation to the card statuses was done based on the statuses available on the DWH, and the mapping of those to VM states?
I think PoweringUp and PoweringDown should be mapped to Up.
Based on Comment #3 and Comment #8, moving PoweringUp and PoweringDown to *Up* and ImageIllegal to *Warning*. The full maps is: Up = PoweringUp, Up, SavingState, RestoringState, PoweringDown Warning = Paused, MigratingFrom, MigratingTo, Unknown, WaitForLaunch, Suspended, ImageIllegal Down = Down, NotResponding, RebootInProgress, ImageLocked
Verified in rhevm-4.1.1.2-0.1.el7.noarch ovirt-engine-dashboard-1.1.0-5.el7ev.x86_64 Powering up/down VM is now reported in Dashboard inventory card as Up. I can also confirm the Dashboard <-> VM status map mentioned in comment 9 by Scott. Dashboard status -> VM search query: Up -> Vm: status = poweringup or status = up or status = savingstate or status = restoringstate or status = poweringdown Warning -> Vm: status = paused or status = migratingfrom or status = migratingto or status = unknown or status = waitforlaunch or status = suspended or status = imageillegal Down -> Vm: status = down or status = notresponding or status = rebootinprogress or status = imagelocked