mirror of
https://github.com/harvester/harvester-ui-extension.git
synced 2026-11-21 16:22:08 +08:00
d8d33ff14f6ece9b31d24990326282367f9df25f
872
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
|
d8d33ff14f |
feat: page the VM list on the server when the API supports it (#1261) (#1273)
* feat: page the VM list on the server when the API supports it
On clusters whose API serves lists from the Steve SQL cache, the VM list
uses PaginatedResourceTable: it loads one page of VMs, and the VMIs, pods,
PVCs, PVs, migrations, restores and backups of those VMs only. Sorting,
search and the namespace filter run on the server, so CPU, memory, IP and
node can't be sorted or searched there.
The feature is gated by the vmListPagination feature flag and enabled for
the VM list only, through a pagination context. Older clusters keep the
current list.
* feat: enable the VM list pagination from v1.10.0
The backend change that serves API lists from the SQL cache has to be on
the v1.10 branch before this is merged.
* fix: keep the paged VM list when adding a volume
The Add Volume dialog loaded every VM to find the volumes in use, which
replaced the page the VM list keeps in the store and emptied the list.
When the list is paged, the dialog now reads only the VMs of the
namespace and doesn't store them.
The requests for the resources of a page use one IN filter instead of
one filter per value, which keeps the URLs about a third as long.
* fix: keep the paged VM list intact in its dialogs and on older APIs
The Add Volume, Storage Migration and Restore dialogs read the VMs, PVCs
and backups of the VM's namespace through the VM. While the list keeps a
page of a type in the store they read the namespace without storing it,
so the list keeps its page and the dialogs' own lists don't shrink when
the page reloads. Storage Migration no longer offers volumes of VMs on
other pages as targets.
The list is paged only when the API lists the VM's printer columns as
metadata.fields, which the SQL cache does, so a missing or non-semver
server version can't turn it on against an API that can't filter.
One failed request for a page's resources no longer stops the others.
* fix: show the restart notice only when its VMs change
The list rebuilds the names of the VMs that need a restart on every
refresh, and when it pages also on every page, sort and search, so the
notice popped up again each time even when nothing new needed a
restart. Show it only when the set of VMs changes.
While paging it still covers only the VMs on the current page: the API
can't filter VMs by their RestartRequired condition.
* fix: find the launcher pods of every VM on the page
The paged list loaded the page's launcher pods by the harvesterhci.io/vmName
label, which only the Harvester VM form writes into a VM's template. VMs
created any other way, like an imported VM or one applied with kubectl,
got no pod, so their rows lost View Logs and the starting and pod error
states.
Load the pods after the page's VMIs instead, by the kubevirt.io/created-by
label KubeVirt puts on every launcher pod (the VMI's UID).
Also cover the paged mode in the tests: it must not load every VMI,
migration or pod, and a forced refresh must reach every request.
* fix: notify about restarts again when coming back to the VM list
The restart notice compared the VMs that need a restart with the
previous value of the list. Coming back from a page that loads every VM,
like Volumes or a VM's detail page, that previous value already held the
same VMs, so the notice didn't show. Remember what was notified on the
list itself instead, so each visit starts fresh, and compare by namespace
and name, so a VM of the same name in another namespace counts as a
change.
Also say which VMs lack Harvester's vmName label: the ones from Forklift
or plain manifests, not every VM created outside the UI.
* refactor: keep the VM list's paging helpers in the component
Only canPaginate stays outside, because ResourceList calls
$loadingResources(route, store) on the component options, where there is
no `this`. The pagination headers become a computed property and the
printer column lookup a method.
* refactor: always list VMs with PaginatedResourceTable
PaginatedResourceTable handles both modes. fetchSecondaryResources loads
every pod, volume, restore, backup, VMI and migration when the list isn't
paged, and only what every row shares when it is. VMIs without a VM are
added back through localFilter, which only runs without pagination.
The list passes the pagination context only when the cluster's API can
page, so older clusters keep the full list.
* refactor: register the VM list's columns with the product
Both column sets go through the product's headers(), with the paged set
next to the unpaged one in config/table-headers.js. The list only adds the
node and restore progress columns, which the API can't sort or search when
paging. State sorts by the VM's Status printer column, metadata.fields.2.
* fix: show the restart notice when coming back to the VM list
Since the list always uses PaginatedResourceTable it reads the VMs from
the store. Coming back to the list the store already holds them, so the
restart watcher never saw a change and the notice didn't show. Run it
when the list is created too.
While paging, the store can still hold the VMs another view loaded, like
every VM after Volumes or a VM's detail, so wait for the list's own page
before notifying.
* fix: let the namespace filter know the VM list is paged
The VMs were registered for pagination with a context that only the VM
list passed. The header's namespace filter asks without one, so it didn't
know the list was paged and kept offering "Not in a Project", which the
paged API can't filter by: picking it hid every VM.
Register the VMs in every context and answer per cluster in the store's
paginationEnabled getter instead: only when the feature is on and the
API lists the VM's printer columns as metadata.fields, as the SQL cache
does. The filter, the list and the store now agree, and clusters whose
API can't page keep the full list.
The filter is shared by every list, so Not in a Project can still be
picked on a list that doesn't page. The paged VM list then asks for
another option, like the shell's generic lists do, instead of showing
no rows.
* fix: keep the paged VM list when showing a VM's configuration
The Config tab of the Show Configuration drawer renders the VM form,
which loads every VM, PVC, PV, VMI and migration, and its device tabs
every VM again. On a paged list that replaced the list's page in the
store and emptied the list.
Hide the Config tab while the store keeps a page of VMs, so the drawer
shows the YAML. The VM's own page still shows its configuration, and a
drawer opened where every VM is loaded keeps its Config tab.
* refactor: define the VM list's node and restore columns with the others
The node and restore progress columns are now in VM_HEADERS, so the paged
set turns their sort and search off through VM_PAGINATION_FIELDS, like the
other columns the API can't index. The list only leaves them out when they
don't apply: the node column when the cluster's nodes can't be seen, the
restore progress while no backup is being restored.
---------
(cherry picked from commit
|
||
|
|
95921c85cd |
feat: add the Kube-OVN VPC Egress Gateways page (#1263) (#1269)
* feat: add the Kube-OVN VPC Egress Gateways page
List, create and edit VpcEgressGateway resources under Overlay Networks >
NAT & Internet, right below the NAT gateways, which are now labeled NAT
Gateways. The form covers the VPC, replicas, internal and external
subnets, SNAT policies, internal and external IPs, BFD and the bandwidth
limit, and keeps the fields it does not show.
* feat: add the VPC BFD port and link the Egress Gateway BFD warning to it
The VPC form gets a BFD Port tab with the enabled switch and its IP,
which an Egress Gateway with BFD needs. The gateway form now checks the
VPC's spec.bfdPort instead of a spec.bfd field that VPCs don't have, and
its warning links to the VPC. Bandwidth also takes quantities such as
1Gi, and a gateway that only uses selectors no longer needs a policy.
* feat: check the Egress Gateway and BFD port against what Kube-OVN accepts
Kube-OVN runs as a secondary CNI in Harvester, so an Egress Gateway's
internal subnet has to sit on a virtual machine network and its external
subnet on a network attachment; the subnet lists now offer only those.
The internal and external IPs must be unique and inside their subnet,
other than its gateway, and the two subnets must differ. Selectors don't
work in that mode, so a gateway needs a policy again. Policies set with
IP blocks show them, and BFD intervals hidden while BFD is off fall back
to their defaults.
Save stays disabled while either form is invalid, each field shows its
own error, and a tab with errors is marked. The VPC's BFD port IP can't
be in one of its subnets, its tab warns when it is turned off while
gateways of the VPC use BFD, and the gateway's BFD warning opens the
VPC's BFD Port tab in a new tab.
* fix: show why Save is disabled on the Egress Gateway and VPC forms
Every field that can block Save shows its own message once touched:
the name, the VPC and both subnets, which have none when left empty.
When the chosen VPC has no subnet on a virtual machine network, the
Internal Subnet placeholder says so.
A subnet can only be in one policy: the gateway leaves a source out of
SNAT when any policy has SNAT off for it, so the same subnet in a second
policy would silently ignore one of them. Enable SNAT explains what
turning it off needs.
* fix: keep one Egress Gateway policy for each SNAT setting
Kube-OVN merges the sources of every policy with the same SNAT setting,
so a second policy with SNAT on (or off) only adds to the first one.
The form now flags it on the second policy, Add Policy adds the setting
that isn't used yet and goes away once both exist, and the Policies tab
says a gateway has at most two policies. Policies with IP blocks are
left out of the check, since the form can't move those.
* refactor: ask for the Egress Gateway subnets with and without SNAT
Kube-OVN puts the sources of every policy into two sets, with SNAT and
without, and leaves a source out of SNAT when any policy has SNAT off
for it. A list of policies with a checkbox each let the form build
policies that only repeated or silently overrode each other, so the
Policies tab now asks for those two sets: Subnets with SNAT and Subnets
without SNAT, a subnet in one list only and at least one subnet in all.
Saving writes one policy for each list that has subnets.
Editing reads existing policies the way Kube-OVN applies them, so a
subnet that was in both lands in Subnets without SNAT, and IP blocks
set outside the form are kept and shown read-only.
* refactor: go back to the Egress Gateway policy list
The VpcEgressGateway CRD and the issue describe policies as a list,
each with snat and subnets, and Kube-OVN accepts any number of them:
policies with the same SNAT setting just work together. The two subnet
lists hid that shape and rewrote existing policies, and they couldn't
mark "one of these two" as required. This puts back the policy list,
where each policy needs a subnet and a subnet can be in one policy
only, which is the one case that changes what the gateway does. The
Policies tab now says how policies work together.
* refactor: ask for the Egress Gateway subnets with and without SNAT again
Kube-OVN only uses which subnets have SNAT and which don't, so two
lists, Subnets with SNAT and Subnets without SNAT, hold the same data
as the policy list, are clearer to fill in, and can't build a policy
that repeats or overrides another. Saving writes one policy for each
list that has subnets. Neither list is marked required: the Policies
tab says one subnet in either list is needed, and both lists show that
message once touched while they're empty.
* fix: show the Egress Gateway policy rule once for both subnet lists
One subnet in either list is enough, but a warning on each list read as
if both lists were required. The lists no longer show a warning; one
message under them says to select at least one subnet in either list
while both are empty, next to the tab mark and the disabled Save.
* fix: say why the Egress Gateway subnet lists are empty
The policy lists only offer subnets of the VPC chosen on the Basic tab,
so before a VPC is chosen they were empty with no reason given. They
now say to select a VPC on the Basic tab first, or that the chosen VPC
has no subnets.
* fix: make the Egress Gateway fields that need a VPC say so
Internal Subnet and both policy lists only offer subnets of the chosen
VPC, so they're disabled until a VPC is chosen and say to select one
first. The policy lists keep their dropdown attached to the field, so
it follows the field when a chip is removed instead of leaving a gap.
The BFD intervals are labeled Receive Interval and Transmit Interval,
which fit next to their unit, with a tooltip each.
* fix: keep the external subnet out of Egress Gateway policies and check usable IPs
Kube-OVN reroutes a policy's sources to the gateway, so the gateway's
own external subnet as a source would send its outbound traffic back
into it; the policy lists no longer offer it and the form flags it.
Static IPs must also be usable addresses: Kube-OVN assigns from the
network address + 1 to the broadcast address - 1, so those two are
rejected along with the gateway.
The policy rule is an error banner like the IP count rule, the BFD
multiplier has its own message, the IP Addresses and Bandwidth
instructions are hidden on the read-only page, and the VPC's warning
about gateways that need its BFD port reads right for one gateway.
* chore: declare ipaddr.js, which the subnet model imports
It was only installed as a dependency of webpack-dev-server.
* feat: support PrimaryCNI Egress Gateways
Kube-OVN v1.17 can send a gateway's traffic out through the pod network
when it isn't the primary CNI, as on Harvester. Such a gateway takes no
external subnet, external IPs or bandwidth, so the form couldn't save one.
The Basic tab now picks the external network, locked after creation, and
hides the fields PrimaryCNI doesn't take.
* fix: offer Primary CNI only when the cluster's CRD has it
Kube-OVN v1.16, in Harvester v1.10.0-rc1 and older, has no
externalNetworkMode and requires an external subnet, so a Primary CNI
gateway was rejected there. The form reads the CRD's schema definition
and only shows the External Network choice when the field exists.
---------
(cherry picked from commit
|
||
|
|
e7e1a937d5 |
feat: add persistent VM state storage guidance (#1227) (#1267)
When Longhorn installation is deferred, vmstate-persistence StorageClass is not created. Add prerequisite guidance in the VM form:
- Disable persistent TPM and EFI options when vmstate-persistence is unavailable
- Create mode: clear selections and show banner explaining how to add
Longhorn disk or create vmstate-persistence SC manually
- Edit mode: preserve existing values, keep options disabled, show banner
- Hide banner when vmstate-persistence exists or in View mode
(cherry picked from commit
|
||
|
|
8ac059b2d6 |
ui: add Upgrade Storage Class selector for v1.10+ upgrade flows (#1249) (#1266)
Add a StorageClass selector to online and air-gapped upgrade pages when
the repositoryStorageClass feature is enabled (v1.10+).
- Online: stores selection in harvesterhci.io/upgrade-storage-class annotation
- Air-gapped: stores selection in image spec.targetStorageClassName
- Selector locked while uploaded image exists; unlocked on switch to URL
- Hidden on pre-v1.10 clusters
(cherry picked from commit
|
||
|
|
ff0a6c8431 |
fix: bind bulk VM disk deletion checkbox to checkAll (#1146) (#1264)
* fix: bind bulk VM disk deletion checkbox to checkAll
* test: cover the VM delete prompt checkboxes
---------
(cherry picked from commit
|
||
|
|
38473b1e90 |
feat: make the ignore rule dialog and the ignored checks clearer (#1255) (#1260)
* feat: make the ignore rule dialog and the ignored checks clearer
In the ignore rule dialog, "Ignore for" becomes "Expiration" and
"Entire check" becomes "All resources". Editing an expired rule now
opens with a warning at the top that says when it expired, and only
its expiration can change until a new one is set. The warning that
the rule stays expired, shown at the bottom after a change, is gone.
In the Components table, "+ N ignored" is a link like "+ N more" and
shows that check's ignored resources. The checkbox only brings back the
checks ignored as a whole, reads "Show ignored checks" and counts them
in a badge. Checks with anything ignored get the warning banner's
background, and the filter also finds ignored resources.
* fix: put Expiration first and keep a locked Scope from changing
An expired rule's fields wait for a new expiration, so Expiration is
now the first field of the dialog instead of coming after the fields it
unlocks.
The shell's RadioGroup still takes the Up and Down keys once it is
clicked, even when it is disabled or in view mode. So on an expired rule
a click and Up switched the locked Scope, and Save then widened a rule
for some resources to the whole check. The dialog now ignores a scope
change while the rule is locked or only viewed.
---------
(cherry picked from commit
|
||
|
|
47a9de7661 |
feat: redesign the VNC console (#1233) (#1254)
* chore: bump @rancher/shell to 3.0.15-rc.1 for the Rc components
* feat: rebuild the VNC console toolbar with Rc components
Name and namespace, a connection status pill with its own menu, Rc tertiary buttons with the design's icons, and the keyboard hint under the screen. noVNC's grey backdrop is gone and the screen keeps a thin frame that turns into the keyboard-focus ring.
* feat: move VNC shortcut keys into one Rc dropdown menu
Saved shortcuts, Ctrl and Alt toggles that apply to the next key, Print Screen, F1 to F12 and the way into custom shortcuts now share the Shortcut keys menu, replacing the cascading popover and the separate custom keys menu. Saved shortcuts can carry a name; the old list format still reads. Focus goes back to the screen after a key is sent.
* feat: replace the VNC settings popover with the View options panel
Automatic or Manual scaling on the shell ButtonGroup, a zoom slider with step buttons, JPEG quality and compression with their trade-off spelled out, and Reset to defaults. The screen now scales to fit by default, inside its focus ring. View only is gone with the old panel.
* feat: confirm Ctrl+Alt+Del and Soft reboot in an RcModal
Small RcModal with the VM's name in the title and a destructive primary button, coloured the way the shell's own dialogs do it. Esc and the background now close it too, and the keyboard goes back to the screen afterwards.
* feat: rebuild VNC custom shortcuts as an RcModal with a key recorder
Record a combination by pressing it, name it, and see the saved ones as RcTags with a delete button each. Every change is saved straight away. Keys keep the format older versions read; names live in their own preference so those versions keep working.
* feat: show VNC connecting and connection failures as console overlays
A spinner with the VM's name while connecting (and the attempt count while retrying), and a card with an icon badge, title, explanation and Rc buttons when the console cannot be used: in use, unavailable, connection lost or failed. Take over confirms in an RcModal.
* feat: float a compact VNC bar over the screen in full screen
Name, connection status and Exit full screen in a floating bar instead of the toolbar, with the keyboard on the screen. Take over leaves full screen first, since its modal opens outside the console.
* fix: VNC console review fixes
Show a not-running state instead of connecting forever when the VM has no instance, and find the VM by its uid so the name shows. Keep a failed shortcut save from looking saved. Offer Take over only when another session holds the console, and Reconnect now while retrying. Wrap the toolbar on narrow windows, keep the screen black so overlays read in the light theme, keep the destructive button red on focus, announce only the status label, let Esc and Tab leave the key recorder, release held modifiers when the menu closes, and let Ctrl+T and Ctrl+N through in full screen.
* feat: use the redrawn VNC console icons
Power, maximize, minimize and wifi-off on the design's 24px grid, the 20px connecting spinner, and a status card badge that stays dark on the black screen. The screen is only black while there is no live picture.
* feat: run standalone Harvester Prime with the SUSE brand
Standalone Harvester reports ui-brand=harvester, which has no stylesheet, so Prime only got a green primary on top of the default palette: 33 of the 36 SUSE theme tokens differed from Rancher Prime (blue tertiary buttons, another green, dark text on primary buttons, no SUSE font). On the Prime flavor the brand is now switched to suse in this browser, and all 36 match Rancher Prime in light and dark.
* fix: name recorded VNC shortcut letters after the key pressed
macOS Option turns letters into symbols (Option+G gives ©), so a recorded Alt+Cmd+G was labelled and sent as ©. Letter keys now take their name from the key code when recording, labelling and sending, which also fixes shortcuts saved before.
* fix: give the VNC shortcut menu chips the secondary button colours
The key chips, held modifiers and the manage item used the link colour, so they stayed blue under the SUSE brand. They now take RcButton's secondary, active and tertiary tokens and follow the brand like the toolbar buttons.
* fix: VNC console review fixes, round three
Wait for the instance to run before connecting and retry refusals right after a start, keep Reconnect now counting retries, stop key sending once the VM is down, only rename option-mangled letters (Alt without Ctrl, dead keys) so other layouts keep their keys, give the keyboard back to View when its panel closes, add a shortcut with Enter, keep long shortcut names to two lines, only replace Harvester's own brand and refresh the favicon, and stop relying on stylesheet order for the spinner and danger button.
* feat: change the VNC menu keys with the held modifiers
The keys under Hold modifiers follow what is held, so the menu sends what the old cascading menu did: Ctrl+N/T/W/Y with Ctrl, Ctrl+Alt+Del with Ctrl and Alt (through the same confirmation as its button), and Print Screen and F1-F12 alone or with Alt. Ctrl+Alt+F1-F12 stays for switching Linux consoles. The recorder points to the menu for browser-reserved combinations.
* feat: bring View only back to the VNC View menu
The View menu has a View only switch again, as the old settings panel did: the screen stays live but no mouse or keyboard input reaches the VM, the status reads Connected (view only), the key menus are disabled and the hint says how to turn it off. The design does not have it yet.
* feat: move View only into the VNC connection menu
View only is how this session uses the connection, so it sits in the Connected menu (also in full screen) as a checkbox item, and the View menu is only about how the screen is shown.
* feat: keep the VNC console's controls usable around the connection
Full screen goes over the whole page and its bar gets Shortcut keys and Ctrl+Alt+Del, so menus and the confirmation show there too and Take over no longer has to leave full screen. Soft reboot works whatever the console's state, since it does not go through it. The Shortcut keys menu always opens: while keys cannot be sent it says why and disables only the keys, so custom shortcuts can still be managed. The focus ring now sits 2px off the screen so it shows against a dark framebuffer edge in the light theme.
* feat: disable the VNC console's input controls while View only is on
View only sends nothing to the VM, so Shortcut keys (managing shortcuts included), Ctrl+Alt+Del and Soft reboot are disabled instead of the menu opening to explain why. The menu still opens while disconnected, as there are shortcuts to manage. The hint under the screen takes the new wording, read raw since t() would escape its quotes.
* feat: fill the window with the VNC console in full screen
The screen takes the whole window, with no padding, ring or inset, and black around a picture of another shape. It always fits in full screen, whatever View mode is saved, which comes back on leaving. The bar floats over the screen, slightly transparent until it is hovered, holds the keyboard focus or has a menu open, and no longer has Ctrl+Alt+Del, which stays in the Shortcut keys menu.
* feat: show the mouse over the VNC console
noVNC hides the mouse over the screen until the guest sends a pointer image, and a text console or a guest that draws its own pointer never does. Show the system pointer unless one arrives, so it can be seen in the window and in full screen. A guest that sends an image keeps it.
* fix: close the VNC console's menus on a click on the screen in Firefox
Firefox delivers the click to the canvas, where noVNC stops it, so an open menu never saw a click outside it. Full screen makes this matter, with little else left to click. noVNC does nothing else with a click, so let it through.
* feat: make View only a toggle in the VNC console toolbar
View only moves from the Connected menu to the shared ToggleSwitch between the connection status and Shortcut keys, in the toolbar and in the full screen bar, so its state is always in view and the status no longer has to repeat it. Turning it off hands the keyboard back to the console. The full screen bar no longer gets only half the window from its left: 50%, which would have squeezed its buttons with the wider content; it wraps when it really does not fit.
* feat: leave View only out of the VNC console's full screen bar
View only is set from the toolbar. A session that enters full screen with it on stays that way, with Shortcut keys disabled, and changes it after leaving.
* feat: make View only a checkbox in the VNC console toolbar
It takes the shared Checkbox instead of ToggleSwitch, in the same place between the connection status and Shortcut keys. Clicking its label now turns it off as well as on.
* fix: save the VNC view settings only when the user changes them
The previous console wrote its defaults to localStorage the first time it
opened, so Manual zoom at 100% looked like a choice and the new Automatic
default never reached anyone who had used the console. Settings now live
under a new key and are saved only when the user changes them.
* fix: keep View only on when the VNC View panel resets its settings
View only moved to the toolbar, but Reset to defaults in the View panel
still set it back to off, so the screen started sending the mouse and
keyboard to the VM again. The reset now leaves View only as it is.
* fix: give the VNC quality and compression sliders their full height
The slider rule's flex: 1 is meant for the zoom row. In the quality and
compression sections, which are columns, it set the height instead and
left those two sliders 4px tall, so a click on the visible thumb missed.
* fix: reach the VNC shortcut keys from the keyboard after holding a modifier
RcDropdown collects its items when the menu opens, and holding Ctrl or
Alt swaps the keys below the toggles while it is open, so the arrow keys
stopped at Alt. The menu now collects its items again on each key press,
as RcDropdown does for its submenus.
* fix: show a failed custom shortcut save inside the dialog
The dialog only reported a failed save with a growl, and the console
page's layout has no growls, so a delete or an add that could not be
saved changed nothing and said nothing. The error now shows as an alert
in the dialog until the next change or until it closes.
* fix: keep the focus in the custom shortcuts dialog after adding or deleting
The delete buttons and Add shortcut were disabled while a change saved,
and disabling the button that was just pressed dropped the keyboard's
focus to the page. They now stay enabled, and a change made while one saves is
ignored. After a delete the next row's button takes the focus, or the last
one's, and after an add, or once the list is empty, the recorder does.
* fix: record a modifier pressed on its own as a custom VNC shortcut
The recorder held every modifier for the key that ends the combination,
so the Windows key on its own could not be saved any more, while the old
dialog could. Modifiers released before any other key are now the
combination, once the last of them is released.
* fix: stop the custom shortcut form saying a new shortcut is already saved
The list shows a new shortcut while it saves, and the form kept it until
the save came back, so every add showed "This key combination is already
saved." as an alert for as long as the save took. The form now empties
as soon as Add shortcut is pressed, and the shortcut goes back in it if
the save fails, unless the next one was started meanwhile.
* fix: make the VNC Reconnect button an RcButton
The plain button took the global 40px button height inside the 32px
Disconnected pill, so its focus ring stuck out above and below the pill.
As a small ghost RcButton it is 24px tall and its ring stays inside.
* fix: carry View only over from the previous VNC console
The new storage key dropped a View only that someone turned on in the
previous console, so their mouse and keyboard reached the VM again
after the upgrade. View only was off unless the user turned it on, so
it is read from the previous key until this console saves its own
settings; the display settings still start fresh.
The comment on the key now says the settings are written once the user
changes one, rather than only the ones changed.
* fix: warn that a VNC soft reboot loses unsaved work
The soft reboot confirmation said the console disconnects while the VM
restarts and reconnects when it is back, which it does not: a soft
reboot restarts the guest in the same VM instance and pod, so the
console stays connected. It also dropped the previous dialog's warning
that unsaved work is lost, which is back.
---------
(cherry picked from commit
|
||
|
|
96629e92f5 |
chore(deps): align resolutions and yarn.lock with @rancher/shell 3.0.14 (#1239) (#1240)
* chore(deps): align resolutions and yarn.lock with @rancher/shell 3.0.14
yarn only applies resolutions from the root package.json, so the ones
@rancher/shell 3.0.14 declares never reached the extension. Copy them
over, use the **/ form for diff2html/diff since diff2html sits under
@rancher/shell here, and drop the cronstrue pin so the extension and the
shell both use the 3.9.0 the shell depends on.
Re-resolve yarn.lock against the shell's own lockfile so every shared
package lands on the version the shell ships, and semver and
@types/node follow the shell's versions too.
* chore(deps): keep the cronstrue pin
cronstrue 3 describes @reboot, so the backup schedule form enabled Create
for it while Harvester's webhook rejects it. The dashboard itself runs
cronstrue 2.53.0 and the shell went back to that after 3.0.14, so keep
our 2.59.0 pin.
---------
(cherry picked from commit
|
||
|
|
d4a44ae9c6 |
fix: keep the ignore rule name valid when the check has spaces (#1234) (#1238)
Backup schedule checks are named after the schedule's condition reason,
such as "Proactive Schedule Suspend". The rule name used it lowercased,
spaces included, so Kubernetes refused it and the check could not be
ignored. Turn whatever a name can't have into hyphens. The rule keeps
the check as it is in its spec, which is what the controller matches.
(cherry picked from commit
|
||
|
|
fb892c2105 |
fix: show resource counts for all side navigation items (#1196) (#1237)
* fix: show resource counts for all side navigation items
The side navigation looks up the count of a nav item by its name, but the
counts are keyed by the K8s resource type. Many Harvester nav items are
named differently than the resource they list (e.g. harvesterhci.io.volume
lists persistentvolumeclaim), so no count was shown for them. Some list
views also show only a subset of their resource type, so the plain count
would not match the list.
Override the count getter of the harvester store:
- Resolve the nav item to the resource type it lists, using the same
fallback to the configureType resource as schemaFor. This fixes Hosts,
Alertmanager Configs, Cluster Flows, Cluster Outputs, Flows, Outputs
and Secrets.
- Count only the subset shown by the list for Volumes, VM Networks,
VM Backups, VM Snapshots, Volume Snapshots, Cloud Configuration
Templates and Storage. The matches are counted per namespace once per
store list using sharedComputed.
- Like the list views, do not count resources in system namespaces when
"All Namespaces" is selected.
Steve's counts endpoint only returns the total per resource type, so the
filtered counts are computed from the store list and are only available
once that list is loaded. Until then these nav items show no count, unless
there are no resources of that type at all (then 0). The dashboard loads
PVCs and networks, the VM list PVCs and backups, the other resources are
loaded when their list is opened.
Related to: https://github.com/harvester/harvester/issues/11984
* refactor: get the side navigation counts from the server
Stop loading full resource lists and filtering them in the browser to
get the side navigation counts.
- Nav items listing all resources of a type use the counts endpoint,
through the resource type of the nav item.
- VM backups, VM snapshots, volume snapshots, storage classes, cloud
config templates and VM networks request their subset from
/v1/harvester with a page size of 1. These requests run again when the
counts endpoint reports other numbers for the type, and when the
namespace filter changes.
- Namespaces count the already loaded namespaces that are neither system
nor Fleet managed.
- Volumes have no count: the list hides volumes that are only marked by
annotations, which the API cannot filter on.
* refactor: keep the side navigation count helpers in the store getters
Move them out of their own file, next to the count getter that uses them.
* fix: make the side navigation count requests more robust
- Ignore a late response of the previous cluster. Resetting the request
ids when loading a cluster let it overwrite the count of the new one.
- Keep the watchers when a cluster left while it was loading finishes
loading after the new one.
- Show no count instead of 0 on Harvester v1.3, which ignores the query
and sends all resources without a count.
- Hide the storage network by its generated name. Networks created
before Harvester v1.4 do not have the hash label, and an upgrade does
not add it.
- Stop refreshing after leaving the product, cancel pending refreshes
when the watchers stop, and cope with the store module being
unregistered when logging out.
* fix: count only the settings the Settings page shows
The nav item showed the total of all Setting resources, but the page only
shows the allowed ones that exist in the cluster and are supported by its
version.
Move these rules into a visibleSettings getter that both the page and the
count use, so that they always agree. SettingList no longer needs its
feature flag check.
* test: remove a Settings page test that was committed twice
* fix: show the number of volumes in the side navigation
The Volumes list hides golden image volumes and the temporary volumes CDI
uses to populate a volume, so the total of the counts endpoint would be
wrong.
Golden image volumes are only marked by an annotation, which Harvester's
API cannot filter on, but each one belongs to a DataVolume owned by its VM
image. So request the volumes that are not owned by another volume, and
the DataVolumes owned by a VM image, and show the difference.
Every nav item with a list now has a number, so remove NO_COUNT.
---------
(cherry picked from commit
|
||
|
|
2c18f81722 |
feat: align the storage performance options with the Rancher machine pool form (#1212) (#1232)
* feat: align the storage performance options with the Rancher machine pool form
Show the storage performance profile on every volume instead of behind a
collapsible toggle, and move the VM-wide options to a plain Storage
Performance section in the Advanced Options tab. Put the Native I/O tip in
a live region so screen readers announce the cache mode change, use an
integer input for the I/O thread count, and use the same copy as the
Rancher form.
Skip the view-only backup row in the image and container volumes when it
has nothing to show, so it no longer adds an empty gap above the profile.
Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com>
* fix: announce the storage performance warnings and name the virtio requirement
- Put the bus tip and the Filesystem warning in an always-rendered status
region. They appear above the field being changed, so screen readers need
them announced.
- Say in the bus tip that a dedicated I/O thread needs the virtio bus, because
KubeVirt rejects it on any other bus.
- Mark I/O Thread Count as required while editing, like the Rancher form.
Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com>
* fix: state only the dedicated I/O thread requirement in the bus tip
Neither the Harvester docs nor KubeVirt tie cache mode or I/O mode to the bus,
so stop saying that those options work best on virtio.
Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com>
---------
Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com>
(cherry picked from commit
|
||
|
|
94c5c919f5 |
fix: show quotes in high-performance storage hints and keep Custom profile (#1204) (#1270)
* fix: show quotes in high-performance storage hints and keep Custom profile
The Native I/O and I/O threads policy hints were passed to Banner as
:label="t(...)". t() escapes HTML and Banner escapes the label again,
so the quotes showed up as ". Use label-key instead, which renders
the string once.
Picking the Custom performance profile and then choosing options that
match the Default or High Performance preset flipped the selector back
to that preset. Keep an explicit Custom choice and only infer the
profile from the disk values otherwise.
Related: harvester/harvester#11550
* fix: only keep Custom profile when the user picked it
syncProfile() skipped re-detection whenever the profile was Custom,
including a Custom that was inferred from the disk values. In an
existing-volume row, switching from a volume with a WriteBack override
to one with High Performance values left the selector on Custom.
Track the user's explicit Custom choice in its own flag and use that
in the guard, so an inferred Custom still follows the disk values.
Related: harvester/harvester#11550
* test: cover the label-key hints of VMPerformanceOptions
Check that the I/O threads policy and no virtio disk hints are passed to Banner as label keys, so their quotes are not escaped twice, and that the I/O threads policy hint is hidden once a policy is set.
---------
(cherry picked from commit
|
||
|
|
607cb4b8ee |
feat: move the device pages into a Hardware Devices side nav group (#1211) (#1231)
The PCI, SR-IOV Network, SR-IOV GPU, vGPU, vGPU MIG and USB device pages
were listed under Advanced. Give them their own top-level group, placed
after Backup and Snapshots. Add-ons stay under Advanced.
(cherry picked from commit
|
||
|
|
3a432cd996 |
fix: label the VPC group of subnets without a VPC as None (#1210) (#1230)
Kube-OVN sets a subnet's VPC (ovn-cluster when none is given), but a
subnet it hasn't processed, such as one with an invalid provider, has
none. The VPC page grouped those under "Virtual Private Cloud:" with an
empty name, and its Create Subnet, Topology and actions buttons pointed
at no VPC.
Show None as the VPC name for that group and leave out its VPC buttons.
(cherry picked from commit
|
||
|
|
e03d3dbed6 |
fix: block VM creation/save with invalid CloudInit YAML (#1040) (#1229)
* fix(vm): block VM creation with invalid CloudInit YAML content
* fix(vm): add isWindows guard to validateCloudInit in template version
* fix: validate Windows CloudInit YAML on save
* fix: validate cloud-init by format in the VM mixin
Cloud-init picks the user data format from the first line, so scripts,
MIME messages and Jinja templates are valid without being YAML. Parsing
all user data blocked them, and it stopped any existing VM holding one
from being saved, even with nothing changed.
Only parse user data that starts with #cloud-config, and require a
mapping for it and for network data, because plain words are valid YAML
and were still saved.
Both forms already use the VM mixin, so the check moves there and shares
the header pattern with hasCloudConfigComment.
* fix: save other user data formats as typed and check text without a header
The form added #cloud-config to any user data that parses as YAML, so a
script was saved with the header and its lines folded together, and Edit
Config then failed the check on it. User data that starts with a script,
MIME or another cloud-init or Cloudbase-Init format is now saved as typed,
and any other text is checked as cloud-config, with or without the header,
because the form adds the header when saving it.
Editing only checks cloud-init that changed, so a VM saved before this
change can still be edited with the cloud-init it already has.
---------
(cherry picked from commit
|
||
|
|
5e5b01a035 |
fix: show VNC console dialogs in full screen and restore deleting custom shortcut keys (#1222) (#1224)
* fix: show the VNC console's dialogs in full screen
Only the console went full screen and dialogs render outside of it, so the
Ctrl+Alt+Del and Soft Reboot confirmations, among others, stayed hidden behind
it until full screen was left. Put the whole page in full screen instead.
* fix: bring back deleting saved VNC console shortcut keys
The Vue 3 migration dropped the close button from the list of saved shortcuts,
which left no way to delete one. Restore it, and remove by position in the list:
the old code parsed the displayed HTML to find the shortcut, which cannot match
one with the Space key in it, such as Ctrl+Space.
* fix: tidy the VNC console's custom shortcut keys
Give the saved shortcuts in the menu the same space as the separator above
them, and name the dialog's buttons Cancel and Save, which say what they do.
---------
(cherry picked from commit
|
||
|
|
0a797e7e22 |
feat: add shutdownVMOnMigrationFailure to the upgrade-config setting (#1218) (#1265)
* feat: add shutdownVMOnMigrationFailure to the upgrade-config setting Add a control for the new field, which decides whether a VM that fails to live migrate during the node pre-drain phase is shut down so that the upgrade can continue. The field is only shown and defaulted on v1.10.0 or later clusters. * fix: name the upgrade-config radio groups for screen readers Pass the labels of the Restore VM and Shut Down VM on Migration Failure radio groups as their aria-label, so that screen readers can tell them apart. Select the value through the radio group in the tests, so that they catch a group that isn't bound to its field. --------- Signed-off-by: Tim Liou <tim.liou@suse.com> Co-authored-by: Tim Liou <tim.liou@suse.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
ed280e2f81 |
feat: add resource health dashboard in a Components tab (#1225) (#1228)
* feat: add component health dashboard sections
Add health summary and component health tables with severity filters and expandable check details.
* feat: use health API group for component health resources
Update dashboard resource types to use the health.harvesterhci.io API group.
* feat: add component health ignore controls
* feat: gate component health behind the componentHealth feature flag
Only show the component health dashboard sections on Harvester v1.10.0 and later.
* fix: address component health review comments
- sort the health summary table by a severity rank
- use buttons instead of anchors for restore and show more actions
- let the affected resources list expand so every resource can be restored
- widen the ignore popup and cap its height
* refactor: split component health into dedicated components
Move the health summary, component health tabs and ignore popup out of
the dashboard page into components/health, and share the constants and
helpers through utils/component-health.
* fix: sort component health by severity rank
* fix: toggle health rows in place and tell empty filters from no checks
- toggle the expanded message and resource ids by mutating the unwrapped
arrays, the template does not pass refs
- show "No checks match the selected filters" when the severity filters
hide existing checks, keep "No issues detected" for a report without
checks
* refactor: fetch component health resources in their own components
* feat: move health into a Components tab with the checks as sub-rows
The dashboard showed the summary table, then side tabs with a second
table for each component. It is now one table in a new Components tab
next to the Events tabs: one row per component, which opens to its
checks. Rows start closed, and the component name is a button so the
keyboard can open them. The tab only shows when the addon's resources
exist.
The table's own filter box replaces the severity dropdown. It only looks
at the table's own rows and the checks are in the sub-rows, so the rows
give it the text of their checks, a row only shows the checks it found,
and the rows it found open.
Ignoring a check moves from a popover, which cut off its Reason field
and Ignore button, to an RcModal that starts over each time it opens.
* refactor: use RcButton for the component health buttons
The Ignore modal's button already was one. This moves the rest of them,
the row's Ignore, Restore, the "show more" links and the component name,
off the global btn classes.
* refactor: manage one ignore rule per check from the row's actions
A check now has one rule, reached through the table's action menu: create
it when the check has none, view, edit or delete it when it has one. View
and edit use the same modal as create. This replaces the Ignore button and
the Restore links, which could only delete a whole rule and said the check
was restored when it wasn't.
The modal's section titles were the radio group's h3, as big as the modal's
own title. They follow what rancher/dashboard does in its forms now.
The actions follow what the user may do. The rules are only loaded with
their schema, and a list that fails no longer empties the table.
* refactor: let the tables filter by themselves and show resources as chips
The filter had logic of its own on top of the table's. The nested table now
filters its rows the way the outer one does, with what is typed in the box.
Rows no longer open by themselves, which is how the other tables with
sub-rows behave. Expand All and Collapse All are there instead, copied from
what the table has from shell 3.0.15.
Affected resources are chips under the name the UI has for their type, with
the usual counter badge and the API version beside it. Ignored resources are
outlined instead of filled, and the text of an ignored check is muted. The
links use the shell's wording.
The component rows lose Last Updated. The summary only has one time, for its
counts as a whole, so it said nothing about the component in the row. They say
Error instead of Critical, as the checks inside them do. The Node column goes,
as nothing sets its label.
The Components tab shows the number of errors as a badge, as nothing else on
the page tells there are any.
* fix: name the ignore rule after its check and make its expiry clear
- Name the rule the UI makes after its check, so Kubernetes refuses a
second one for the same check. The row's menu only manages that rule.
- Show the rule's current expiry above the durations, and the new one
under them once a duration is chosen. Edit starts with none chosen,
Save stays off until something changes, and a warning says when an
expired rule would stay expired.
- Delete the rule through the shell's delete prompt, like every other
delete.
- Say which list could not be loaded, instead of showing it as empty.
- Keep the full count of affected resources, say how many ignored ones
are left out of the list, and count them on Show ignored.
- Drop the checks' Last Updated, which is the time of their whole report.
- Leave out the row menu when no check has an action.
- Mark the Components tab with the shell's alert icon when there are
errors, with their number in its tooltip, instead of a badge.
---------
(cherry picked from commit
|
||
|
|
27e6f2940e |
chore: bump version to 1.10.0-rc2 (#1250)
Signed-off-by: Harvester Bot <94133267+harvesterhci-io-github-bot@users.noreply.github.com>v1.10.0-rc2 |
||
|
|
3322162d9a |
chore: bump version to 1.10.0-rc1 (#1213)
Signed-off-by: Harvester Bot <94133267+harvesterhci-io-github-bot@users.noreply.github.com>v1.10.0-rc1 |
||
|
|
6e0e0237ef |
feat: use the SUSE Virtualization product name on Prime (#1202)
* feat: switch the product name with Rancher Prime Use a [[Harvester]] token for the product name in the translations, so Rancher 2.16 fills it from the l10n global that harvester-manager registers: SUSE Virtualization on Rancher Prime, Harvester otherwise. Rancher before 2.16 shows the tokens as written, so on those hosts (and standalone) the extension resolves them to Harvester itself. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * feat: switch the standalone product name with the server flavor Standalone has no Rancher Prime, so register the [[Harvester]] l10n global from the server-flavor setting instead: SUSE Virtualization on prime, Harvester otherwise. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: point the network policy banner to the documentation Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * feat: use the SUSE theme in standalone Prime Standalone Prime kept the harvester brand and only overrode the primary colour, so the rest of the palette and the fonts didn't match Rancher Prime. Switch the ui-brand setting to suse in the store when the server flavor is prime, which makes the shell load the SUSE theme. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: keep the SUSE theme after a brand setting change event The watcher also reads the setting's value, so a change from the server loaded into the same object switches the brand back to suse. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> --------- Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> |
||
|
|
d6f6f7c54c |
chore(deps): bump @rancher/shell to 3.0.14 (#1208)
* chore(deps): bump @rancher/shell to 3.0.14 3.0.14 resolves [[name]] tokens in translations from values registered as l10n globals, which the standalone UI needs to pick the product name. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: save cloud-config template edits with @rancher/shell 3.0.14 The YamlEditor of shell 3.0.14 no longer emits onChanges, so editing a template saved its old cloud config. Copy the text from update:value, which both shell versions emit. Jest transforms @replit/codemirror-vim-core, an ESM-only dependency of the new YamlEditor, like the dashboard does. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: keep the YAML editor heights with CodeMirror 6 The min-height rules targeted the CodeMirror 5 element, so the editors shrank to their text. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: reject IP ranges the network settings don't support The CIDR check of @rancher/shell 3.0.14 also accepts short forms such as 10.1/16, extra parts and IPv6, so the VM migration, RWX and storage network settings saved them. Use a strict IPv4 CIDR check of our own. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: keep the look of the file selector buttons The FileSelector of @rancher/shell 3.0.14 no longer takes button classes and renders a medium secondary button. Pass the variant and size that match the previous primary and secondary buttons. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: keep Enter in a migration wizard field on the same step The Wizard of @rancher/shell 3.0.14 renders its step headers as submit buttons, so Enter in a text field clicked the first one and jumped back to step 1. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: show the created message after a restore, upgrade or upload Since @rancher/shell 3.0.14, save() no longer shows the message for resources with a generated name; callers show it with notifyGeneratedName(). Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * refactor: fold the logging output buffer with the CodeMirror 6 helpers The YAML editor of @rancher/shell 3.0.14 is CodeMirror 6, and its CodeMirror 5 methods are deprecated. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: keep the shell's table views off in standalone @rancher/shell 3.0.14 turns its table views on for every list in the standalone UI, as the server has no ui-configurable-tables feature and the default is on. The table views drop the filters the lists put in the table header, which also left the Images list empty. Add the feature, disabled, to the management store in standalone only. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: reject zero-padded octets in IP ranges Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> --------- Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com>v1.10.0-dev-20261025 |
||
|
|
39ed41e527 |
feat: move OS type, CPU model and system options to the VM Basics tab (#1179)
* feat: move OS type, CPU model and system options to the VM Basics tab OS Type, CPU Model and the CPU/system checkboxes (CPU pinning, USB tablet, TPM, EFI, Secure Boot and their persistent-state options) are core machine settings, but they were buried at the bottom of Advanced Options. Show them on the Basics tab, next to CPU and memory. Install Guest Agent stays in Advanced Options because it changes the cloud config. It now sits right below the User Data editor (the data it modifies) through a new optional slot in the Cloud Config component. Only the placement changes; behaviour and validation are untouched. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: Alejandro Bonilla <abonilla@suse.com> * feat: follow the review on the VM Basics layout - Basics now reads OS Type and CPU Model, then CPU and memory, then the system checkboxes, the CPU Manager banner and the SSH key, in the order proposed in review. The checkboxes sit in one wrapper; USB Tablet drops its extra mt-20 and uses the plain `check` class like the others. - Advanced Options puts Maintenance Strategy on the same row as Run Strategy; Reserved Memory moves to its own row. - The moved labels leave harvester.virtualMachine.advancedOptions: TPM, EFI, Secure Boot and the persistent-state labels are now under harvester.virtualMachine.systemOptions (also used by the template version editor), and the CPU Manager banner text is under harvester.virtualMachine.cpuPinning.cpuManager. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: Alejandro Bonilla <abonilla@suse.com> * feat: match the VM template form and address review on the VM Basics layout The VM template form still had OS Type, CPU Model and the CPU/system checkboxes under Advanced Options, so creating a VM and creating a template put them in different places. Use the same layout as the VM form: - Basics: OS Type and CPU Model, then CPU and memory, then the CPU Pinning, USB Tablet, TPM, EFI and Secure Boot checkboxes (plus the persistent-state options when they apply), then the SSH key. - Advanced Options: Run Strategy and Maintenance Strategy on one row, Reserved Memory on its own row, then Show more and Cloud Config. Install Guest Agent sits right below the User Data editor through the after-user-data slot. Review follow-ups on both forms: - OS Type shows an info icon in create mode warning that selecting an image with an OS label on the Volumes tab may change the OS Type selection, now that OS Type comes before the volumes. - The CPU Manager banner under CPU Pinning is info in create mode, where it only explains why the checkbox is disabled, and stays a warning in edit mode, where it is about the restart. Only placement and presentation change; behaviour and validation are untouched. Signed-off-by: Alejandro Bonilla <abonilla@suse.com> --------- Signed-off-by: Alejandro Bonilla <abonilla@suse.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
2f72f25b6e |
feat: improve VNC console with a version gate, in-use screen and confirmations (#1195)
* feat: improve vnc console Signed-off-by: Andy Lee <andy.lee@suse.com> * fix: keep VNC console retries from kicking the active session Builds on "feat: improve vnc console" (#1147) by Andy Lee. - Keep retrying with preserveSession after a failed reconnect attempt, so a displaced console waits for the other session to be released instead of failing immediately or taking the console back. - Close the RFB socket when noVNC asks for credentials, so a failed console does not hold the VNC session. - Restore Custom Shortcut Keys, which were removed in #1147. - Add unit tests for the console reconnect behavior. Related: harvester/harvester#10420 Signed-off-by: Alejandro Bonilla <abonilla@suse.com> * fix: expose VNC quality and compression settings - Add 0-9 sliders for the JPEG quality and compression level, which were already passed to noVNC but could not be changed. - Mark the connection status as a polite live region so screen readers announce connection changes. Related: harvester/harvester#10420 Signed-off-by: Alejandro Bonilla <abonilla@suse.com> * fix: gate VNC session protection by version and rework the in-use screen - Add the vncPreserveSession feature flag for Harvester v1.7.0, the first release whose KubeVirt (1.6.3) honours preserveSession. Without it the console connects as before, offers no Take Over and does not reconnect on its own after the session drops, so two open consoles no longer keep kicking each other out. - Replace the error page with a short card that says when the console is in use, and confirm Take Over in a modal instead of window.confirm. - Keep the overlay text readable on its dark background. - Show "Connected (view only)", translate the Ctrl+Alt+Del label and remove unused translation keys. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * feat: show whether the VNC console has keyboard focus noVNC hides the canvas focus outline, so nothing showed where typing would go. Draw a ring around the console while it has focus, show a toolbar hint ("Typing goes to the virtual machine" / "Click the console to type"), focus the console when it connects and return focus to it after sending keys from the toolbar. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * feat: confirm Ctrl+Alt+Del and Soft Reboot in the VNC console Both buttons can restart the virtual machine with one click: Soft Reboot always, and Ctrl+Alt+Del on most Linux guests. Ask in a modal first, as the VM list already does for Soft Reboot, and return focus to the console after. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * fix: return focus to the VNC console after its confirm modal closes The modal gave focus back to the toolbar button that opened it, so typing did not reach the console until it was clicked again. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> * refactor: drop the VNC keyboard focus hint from the toolbar The ring around the console already shows when it has keyboard focus. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> --------- Signed-off-by: Andy Lee <andy.lee@suse.com> Signed-off-by: Alejandro Bonilla <abonilla@suse.com> Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> Co-authored-by: Andy Lee <andy.lee@suse.com> Co-authored-by: Alejandro Bonilla <abonilla@suse.com> |
||
|
|
5406c08f76 |
perf: speed up the VM list with many VMs (#1189)
* perf: speed up the VM list with many VMs The VM list scanned whole store lists for every row and re-evaluated the same values per row on every render and sort. With several hundred VMs this blocked the browser on sorting and on every VM update. Performance: - look up launcher pods and migrations through shared indexes (namespace/name -> resource) instead of scanning per row; the indexes are computed per store list, so they stay in sync with in-place store updates - use the same migration lookup (getLatestMigration) in the VM model and on the VM detail page - look up PVCs, PVs, VMIs and restore backups by id - merge VMs and VMIs with a Set - evaluate the VM state once when sorting by state - check the unschedulable condition before the pod based state checks - cache feature flag results and per-schema/per-setting values that are the same for every row Bug fixes: - the launcher pod map was never rebuilt, because the store keeps the same array - pods, volumes and migrations of a VM with the same name in another namespace could be picked up The order of the volumes of a VM (vm.volumes) now follows the VM spec instead of the store order. Nothing depends on it. Measured in a synthetic setup, 100 rows per page, sorting by state (Chrome, dev build): - ~490 VMs, ~500 PVCs/PVs: 1.9 s -> 1.0 s - ~490 VMs, 600 PVs, 200 migrations: 6.0 s -> 0.7 s, main thread blocked during VM updates 78% -> 32% With few volumes the gain is much smaller (8-20%). Related to: https://github.com/harvester/harvester/issues/9513 Signed-off-by: Volker Theile <vtheile@suse.com> * fix: read VM data from the store of the resource instead of a fixed one The migrations of a VMI and the migration badge of the state column were read from the hardcoded `harvester` store. In the Rancher explorer (embedded and standalone) the data is loaded into the `cluster` store, so a running migration was shown neither in the VM state nor as badge. The VM model remembered the store of the current product in a module variable on first use and never updated it. The same model is used in the Harvester product (`harvester` store) and in the Rancher explorer (`cluster` store), and Rancher keeps the page loaded when going from the explorer of a Harvester cluster to Virtualization Management and into the Harvester product. The VM list then read the VMIs, launcher pods, volumes and restores from the wrong store, and running VMs were shown as Starting until the page was reloaded. Read the data from the store the resource lives in (`this.$getters`, in the formatter the store of the current product). In the Harvester product this is still the `harvester` store, so nothing changes there. Not covered: volume details like the lock icon of encrypted volumes. The explorer uses the generic Rancher PVC model, which does not know them, and settings based checks (feature flags, backup target) are not loaded there either. Related to: https://github.com/harvester/harvester/issues/11882 Signed-off-by: Volker Theile <vtheile@suse.com> --------- Signed-off-by: Volker Theile <vtheile@suse.com> |
||
|
|
71141b257a |
feat: Add a unit test target (jest) to the UI extension and run it in CI (#1192)
- Introduce test targets. - Adding additional tests. Related to: https://github.com/harvester/harvester/issues/11892 Signed-off-by: Volker Theile <vtheile@suse.com> |
||
|
|
02862bbeea |
fix: "Failed to resolve component: LabelValue" in the container volume form (#1193)
Related to: https://github.com/harvester/harvester/issues/11893 Signed-off-by: Volker Theile <vtheile@suse.com> |
||
|
|
5f13aadeed |
feat: Make use of status conditions when activating maintenance mode (#1145)
Related to: https://github.com/harvester/harvester/issues/9022 Signed-off-by: Volker Theile <vtheile@suse.com>v1.10.0-dev-20261011 |
||
|
|
da36db143f |
fix: gate high-performance storage on v1.9.1 (#1181)
The storage performance options ship in Harvester v1.9.1, not v1.9.0. Move highPerformanceStorage to the v1.9.1 feature flags so the UI does not offer them on v1.9.0 clusters. Signed-off-by: Alejandro Bonilla <abonilla@suse.com> |
||
|
|
5af181f5b6 |
feat: support network boot (PXE) for VM interfaces (#1176)
Add an "Enable network boot (PXE)" option to bridge network interfaces and a VM-level choice to place network boot devices before or after volumes. The form now owns interface boot order: enabled NICs get sequential boot orders ahead of or behind the volumes, and a stale bootOrder is no longer inherited from the old spec. This also fixes editing a VM whose interface bootOrder was set in YAML (the documented workaround): the form used to renumber disks from 1 while keeping the NIC's bootOrder, producing a duplicate that KubeVirt rejects. The boot order shown on the VM, snapshot and backup detail pages now includes network interfaces. Related: harvester/harvester#3325 Signed-off-by: Alejandro Bonilla <abonilla@suse.com> Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
|
528255dea4 |
feat: expose KubeVirt high-performance storage features in VM and Volume UI (#1133)
* feat: expose KubeVirt high-performance storage features in VM and Volume UI Adds UI to configure KubeVirt high-performance disk/storage settings that were previously only reachable via VMBuilder/Terraform: - Per-volume "Storage Performance Options": a performance profile (Default / High Performance / Custom) exposing per-disk cache, io and dedicatedIOThread on all disk types (VM image/root, new, existing, container) and on the standalone Volume form. - VM-wide "High Performance (I/O Threads and Multi-Queue)": blockMultiQueue and ioThreadsPolicy (+ supplemental-pool thread count), shown only in the VM section since these are domain-level and cannot be set per-PVC. - Compatibility guardrails: io=native forces cache=none; cache=none on Filesystem volumeMode warns; blockMultiQueue requires a virtio disk. Ref: harvester/harvester#11550 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Signed-off-by: Alejandro Bonilla <abonilla@suse.com> * fix: address review feedback on high-performance storage UI Resolves the Copilot review comments on #1133: - Gate the whole feature behind a new `highPerformanceStorage` release feature flag (v1.9.0) so older supported clusters never render the controls or receive unsupported KubeVirt fields. - Force Cache Mode to "none" whenever Native I/O is selected, including from the "Default" ('') cache value, and disable every non-"none" cache option (Default included) while Native is active. Previously the empty cache value slipped past the guardrail and produced an invalid combo. - Clear an already-enabled blockMultiQueue when the last virtio disk is removed, so a disabled checkbox can no longer persist an invalid setting. - Strip stale cache/io/dedicatedIOThread fields from the merged disk spec when a regenerated disk no longer requests them, so switching a disk back to "Default" (or unchecking Dedicated I/O Thread) while editing no longer silently preserves the previous values. - Expose the disclosure toggle's state to assistive tech via aria-expanded and an expand/collapse aria-label. - Add unit tests for the profile transitions and the Native/cache guardrails. Ref: harvester/harvester#11550 Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Signed-off-by: Alejandro Bonilla <abonilla@suse.com> * fix: address second-round review on high-performance storage UI Resolves the two Copilot findings on #1133 and votdev's UX feedback on harvester/harvester#11550: - Restore the disk's original bus when the performance profile goes back to "Default". The "High Performance" preset switches the bus to virtio, but leaving the preset only cleared cache/io/dedicatedIOThread, so a SATA/SCSI disk was silently left on virtio — a bus change that can make an existing guest unbootable. The pre-preset bus is now remembered and put back, unless the user picked a different bus themselves in the meantime. - Pass the VM-wide performance values from the VM detail page to the shared Volume component. The mixin already parsed them via getInitConfig(), but the detail caller never forwarded them, so VMPerformanceOptions fell back to its prop defaults and hid itself in view mode — configured blockMultiQueue / ioThreadsPolicy settings were invisible. They are also refreshed in the value watcher so the panel does not go stale. - Make the VM-wide "High Performance" panel expandable/collapsible, matching the per-volume "Storage Performance Options" disclosure. It stays collapsed unless the VM already has something configured, so the average user is not faced with specialist tuning controls by default. - Fix the unit tests to use Vue Test Utils v2 mount options. They used the v1 top-level `mocks` key, which VTU 2.x ignores, so every case failed on mount. The repo has no jest config or test script, so this was not caught. Add coverage for the bus save/restore behaviour. Ref: harvester/harvester#11550 Co-Authored-By: Claude <noreply@anthropic.com> Signed-off-by: Alejandro Bonilla <abonilla@suse.com> --------- Signed-off-by: Alejandro Bonilla <abonilla@suse.com> Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> |
||
|
|
3c92a06268 |
feat: drop striped LVM volume group type from v1.9.1 (#1167)
The LVM add-on reaches GA in v1.9.1 and only supports dm-thin.
Gate the striped option behind a v1.9.1 feature flag, default new
LVM storage classes to dm-thin, and reset a cloned striped type so
older clusters keep the current behaviour.
Related issue: harvester/harvester#11731
Signed-off-by: Vicente Cheng <vicente.cheng@suse.com>
v1.10.0-dev-20260927
|
||
|
|
6256cdc97a |
ci: pin package version to the git tag for tag builds (#1165)
Sprint tags (vX.Y.Z-dev-YYYYMMDD) only share a prefix with the
x.y.z-dev version in package.json, so every sprint plugin reported
the same version and the catalog image push hit "Cannot overwrite
production image". Bump the tree to the tag before building.
Related issue: harvester/harvester#11047
Signed-off-by: Vicente Cheng <vicente.cheng@suse.com>
|
||
|
|
4323132565 |
deps: update patch dependencies (#1159)
Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> |
||
|
|
d05ec68e16 |
chore: add v1.10.0 and v1.9.1 feature flags (#1156)
Signed-off-by: Andy Lee <andy.lee@suse.com> |
||
|
|
cb2cef5e73 |
fix: avoid userpreferences PUT flood from SideNav refresh trigger (#1119)
kickSideNav toggled a dummy favorite via addFavorite/removeFavorite, which persist through prefs/set and issue a PUT /v1/userpreferences on every toggle, flooding the backend and surfacing 500 errors. Switch to the in-memory prefs/load mutation so the SideNav still re-renders without any network request. Signed-off-by: Andy Lee <andy.lee@suse.com> |
||
|
|
45116b568a |
feat: change logo based on server-flavor setting (#1151)
* feat: use prime logo when server-flavor is prime Signed-off-by: Andy Lee <andy.lee@suse.com> * feat: expose server-flavor setting Signed-off-by: Andy Lee <andy.lee@suse.com> * refactor: copilot review Signed-off-by: Andy Lee <andy.lee@suse.com> --------- Signed-off-by: Andy Lee <andy.lee@suse.com> |
||
|
|
f42d798d10 |
fix: require names for cloud configuration templates (#1138)
Signed-off-by: Victor Chang <t7902195204@gmail.com> |
||
|
|
50b4cf5248 |
feat: add rancher k3k addon UI page (#1137)
* feat: add k3k addon UI Signed-off-by: Andy Lee <andy.lee@suse.com> * feat: add rancher-k3k addon UI Signed-off-by: Andy Lee <andy.lee@suse.com> * refactor: copilot review Signed-off-by: Andy Lee <andy.lee@suse.com> * refactor: copilot review Signed-off-by: Andy Lee <andy.lee@suse.com> * refactor: remove k3k edit page Signed-off-by: Andy Lee <andy.lee@suse.com> * refactor: k3k addon description Signed-off-by: Andy Lee <andy.lee@suse.com> --------- Signed-off-by: Andy Lee <andy.lee@suse.com> |
||
|
|
cba37194b6 |
fix: preserve per-volume storage class when editing image-sourced volumes and fallback to the image StorageClass for new volumes (#1144)
Signed-off-by: Siva Kanakala <siva.kanakala@suse.com> |
||
|
|
bbe0c4643f |
deps: update dependency qs to v6.16.0 [security] (#1134)
Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com> |
||
|
|
00584f1d19 |
chore(deps): bump @rancher/shell to 3.0.13 (#1130)
Clears 26 of the 54 open Dependabot alerts (axios, dompurify, serve-static, and drops vulnerable send / @tootallnate/once copies). 3.0.13 ships the flat-config ESLint 10 toolchain, so lint now routes through the @rancher/shell launcher (keeping the legacy .eslintrc on ESLint 7); a prelint step installs an ESLint-7-compatible eslint-plugin-jest into the isolated toolchain. Signed-off-by: Marcelo Fukumoto <marcelo.fukumoto@suse.com> |
||
|
|
4fa6470343 |
feat: LHv2: graduate from "Experimental" to "Preview" (#1129)
In Harvester v1.9.0, we're moving LHv2 from "Experimental" to "Preview", so I've added a new preview label to the settings interface. I figure given Experimental had the "error" color, I should make "Preview" use the warning color, to make it a bit less alarming :-) Related-to: https://github.com/harvester/harvester/issues/9960 Signed-off-by: Tim Serong <tserong@suse.com> Co-authored-by: Volker Theile <vtheile@suse.com> |
||
|
|
d550046189 |
fix: vGPU doesn't display when creating harvester downstream cluster (#1123)
* fix: vGPU doesn't display in edit VM page Signed-off-by: Andy Lee <andy.lee@suse.com> * fix: align PCI passthrough action buttons height Match the enable/disable passthrough and Actions dropdown buttons to 40px and move the selected-count label below them. Signed-off-by: Andy Lee <andy.lee@suse.com> --------- Signed-off-by: Andy Lee <andy.lee@suse.com> |
||
|
|
5f32a9ade5 |
feat: improve readability for vm migration UI (#1112)
Signed-off-by: pohanhuang <pohan.huang@suse.com> |
||
|
|
3f39d561be |
security: fix 15 transitive Dependabot alerts via yarn.lock re-resolution (#1115)
Re-resolve 15 vulnerable transitive dependencies to their latest in-range patched versions in yarn.lock only (no package.json changes): @babel/core, @babel/plugin-transform-modules-systemjs, brace-expansion, fast-uri, form-data, http-proxy-middleware, immutable, joi, js-cookie, launch-editor, nanoid, shell-quote, svgo, websocket-driver, ws. The remaining alerts are pinned by parent packages (chiefly @rancher/shell) or need a cross-major bump, so they can't be resolved by the lockfile alone; they are tracked separately. Signed-off-by: Marcelo Fukumoto <marceloyfukumoto@gmail.com> |
||
|
|
7c84269ed6 |
fix: exclude forklift namespace from system classification (#1114)
Signed-off-by: Andy Lee <andy.lee@suse.com> |
||
|
|
f63be73b11 | fix: ACCESS_MODE undefined error in forklift mapping page (#1111) | ||
|
|
fabc1cec21 |
fix: prevent duplicated cloud-config for empty user data (#1109)
* fix: prevent duplicated cloud-config for empty user data Signed-off-by: Andy Lee <andy.lee@suse.com> * refactor: AI review Signed-off-by: Andy Lee <andy.lee@suse.com> --------- Signed-off-by: Andy Lee <andy.lee@suse.com> |
||
|
|
ed2b6f2551 |
fix: remove map ownerref writeback and delete maps via plan refs (#1105)
Signed-off-by: Andy Lee <andy.lee@suse.com> |