The reported bug was that the new VM would crash because when clicking through the wizard the pool master is selected by default, not the slave the VM snapshot is on. We already select a default host when a new VM is created from a selected host using HostAncestor, but this property is null for a template (which isn't in the server/VM tree). The fix is that if the selection is a template (which it is when we use new VM from Snapshot) then we instead use its host (using Home()) as the default host. Now the correct host is selected in the VM wizard and the VM creation succeeds.
Signed-off-by: Callum McIntyre <callumiandavid.mcintyre@citrix.com>
The problem here was that TimeSpan.Minutes is always below 60, eg. for 67 minutes it is 7 (with TimeSpan.Hours being 1 then). So our TimeSpan was correct, but if it was between 1 and 2 hours then we would write the minutes in the second hour (which is below 60), ignoring the hour. The fix is to use s.TotalMinutes which for 67 minutes is 67. Hours also had a similar issue, except it would be off by 24 hours for spans between 1 and 2 days.
On the General tab XenCenter shows a list of updates that requires the host to be restarted. This is shown for applied updates for which the required guidance hasn't been done (eg. restartHost or restartAgent after-apply-guidances)
This commit fixes a regression that caused restartToolstack warnings to be not shown in the list of warnings.
The displayed messages are also improved, XenCenter will show Toolstack restart or host restart appropriately and not just restart only as it used to.
Signed-off-by: Gabor Apati-Nagy <gabor.apati-nagy@citrix.com>
We now enumerate pre-Creedence and pre-Clearwater servers once, and check hosts only in those lists for the checks that only apply to those servers.
Signed-off-by: Callum McIntyre <callumiandavid.mcintyre@citrix.com>
Moved from BuildList method to the GraphHelpers class. Since they're only used in this class for this search, and there's similar but broader Regex objects in the Helpers class that are used for parsing, I think it's better to keep them in this class to avoid confusion.
Signed-off-by: Callum McIntyre <callumiandavid.mcintyre@citrix.com>
This avoids unnecessary checks where we are upgrading multiple pools from different versions. Also made the same change for the unrelated pre-Clearwater only checks which had the same issue.
Signed-off-by: Callum McIntyre <callumiandavid.mcintyre@citrix.com>
We now only include the StorageLink check if we have any hosts earlier than Creedence in the RPU.
Signed-off-by: Callum McIntyre <callumiandavid.mcintyre@citrix.com>
We have some new RRDs - read, write, read_latency and write_latency. This adds support for them by adding a friendly name and using it to display them. It also adds some filtering (see GraphHelpers.cs) because for backward compatibility io_throughput_read and io_throughput_write (which differ from read and write only in scale) are still available, but shouldn't be displayed in XC. If the new read or write are available, we filter these old datasources out.
Signed-off-by: Callum McIntyre <callumiandavid.mcintyre@citrix.com>
TickUntilCompletion() is now a static method and again takes an ApplianceAction parameter. It does a null check on each while iteration to handle the case where another thread nulls the action. Also added Run() method to ApplianceAction which initialises the ticker (now a private method), and let child classes call Run() instead.
For other actions we get a ticking timer in the Events view because they use PollToCompletion, which calls their Changed event every 900ms. These actions don't use PollToCompletion, and didn't regularly call their Changed event - so their timer didn't tick regularly. This change adds a lightweight mechanism to call their changed event every 900ms until they complete. This allows the events page timer to tick every second in the same way as it does for other actions.
