If a cluster contains two workloads (e.g. deployments) with the same spec.selector, the button to open a container shell from the workload page or list always picks the first pod that matches this selector without taking namespaces into account.
Fixes https://github.com/rancher/rancher/issues/36344 and https://github.com/rancher/rancher/issues/38506
Signed-off-by: Bastian Hofmann <bashofmann@gmail.com>
- this will never be used, as we'll always need a cluster in the route to fetch the plugin from
- additionally 2 people have hit errors where this caused /home to fail
- if a plugin doesn't override/have a model we fall back on model-loader-require
- by default this is blank (to avoid importing all shell models for all plugins)
- for harvester it's so closely tied by both standard resources and resources in
management/rancher stores that it's hard to manually supply all that we
should just load them anyway
* Make container tabs horizontal.
* Update cronjobs, DaemonSets, Jobs, StatefulSets to use same Workload component.
* Fix 6755 - Wrong sidecar config edit not shown in UI.
* Set Persistent Volume claim capacity as required.
* routing maybe
* remove nested edit views
* harvester custom routes
* fix loading imported dev cluster
* move harvester-manager hci.cluster type out of harvester pkg
* fixes for build-pkg
* fix harvester custom dialogs
* fix harvester pkg routing
* generateDynamicTypeImport hyphenated dirs
rename cloud_credential and machine_config back to cloud-credential and machine-config
* Appease linting
* Fix epinio pkg build
- Includes fixes for some dashboard component references
* Revert "Receiver credential secrets are deleted when secret is set to none"
This reverts commit 836460e2998959bcda0b14a9f6511f248980b7b5.
* Fix plugin modal cancel button in dark mode
* Ensure correct components loaded when a new version of a plugin is loaded
- The route matcher was never updated with new route-->component values
- This was due to `this.router.options.routes` not containing the result of `router.add`
- See https://github.com/vuejs/vue-router/issues/2280
* Revert "Revert "Receiver credential secrets are deleted when secret is set to none""
This reverts commit e524bb3d40ea7adfe35c38ec7870560625965d25.
* fix harvester manager->harvester routing
* rename dialogs
* move harvester-manager files out of harvester pkg
* Revert "remove nested edit views"
This reverts commit e835835943d63e74c50942884fec3706f8703047.
* plugin validators dynamic import - not working
* fix merge conflict
* fix harvester members
* fix conflict
* custom validators
* harvester table formatters
* members routing
* import path fixes
* harvester mutli load
* Port across tweaks made during dynamic plugin work
- harvester plugin can be bundled with the dashboard until dynamic work has been completed
- Nav directly to virt cluster dashboard instead of via /harvester-c-cluster
- Fix missing bottom border on virt cluster list, hide actions (as they're never populated)
* move config/settings and config/harvester-map
fix hci setting custom components
* fix harvester setting import path
* Move over SerialConsole and Novnc, remove harvester.notifications
- These components were so 1-1 with harvester in the end i moved them over
- Also removed the like-for-like generic.notification `harvester.notification` strings
* Remove isSingleVirtualCluster, most of isMultiVirtualCluster
* Fix two nav issues
- harv cluster members --> prefs --> click on-screen back
- dashboard --> prefs --> click on-screen back
Both of these would be better fixed by an incoming change to the default layout (don't show old content before switching page)
* Fix reload of project/namespace page
* Split out project namespace into it's own component (1 or 2)
* Split out project namespace into it's own component (2 or 2)
* use product hideSystemResources opt instead of isVirtual/product===virtual
* fix missing prop in header
* Move FilterLabel and CloudInitType to harv pkg + two other references
* Move Harvester Upgrade components to harv pkg
- Needs a review with Neil regarding generics
* rebase build+load not working
* fix harvester route in authenticated
* Experimental Changes
* Fix merge conflicts
* WIP Fix routing
- includes project/namespace and namespace pages
* Tidying up
* Move harvester core store to plugin, remove harvester code from authenticated flow
- harvester store is a steve store that now lives in the plugin
- harvester `loadVirtual` replaced with a shortened `loadCluster` in it's own store
- Also fix xterm css import
* Fix three routing related bugs
- Fix project/namespace create/edit/cancel/done flows
- Update harvester's loadCluster to better match loadCluster
- Make edit namespace page project resource agnostic
* Build fixes
* Fix create namespace button and other tweaks
* fix two issues with nav out of imported harvester cluster
* Fix alignment of action menu button
- Used in many places
- all generic table action menus
- global settings --> settings
- rke1 cluster detail page machine pool group actions
- project/namespace list project group actions
- resource detail page action menu
* Revert changes to app and pkg creators
- This should fix `check-plugins-build` gate
- It does hide the underlying issue, but that won't have been brought in by this PR?
Co-authored-by: Nancy Butler <42977925+mantis-toboggan-md@users.noreply.github.com>
- harvester plugin can be bundled with the dashboard until dynamic work has been completed
- Nav directly to virt cluster dashboard instead of via /harvester-c-cluster
- Fix missing bottom border on virt cluster list, hide actions (as they're never populated)
- We determine the cluster id via the node's `self` link
- Normally this is something like `k8s/clusters/c-m-274kcrc4/v1/nodes/nodeid`
However for the local cluster this is only v1/nodes/nodeid`
- This meant that for local cluster nodes it's norman and mgmt nodes were unavailable
- From what i can this only affected it's ability to
- determine it's role (though this had a working fallback)
- determine if some actions were shown (and also execute them)
* Ensure that we show empty machine pools
* Fix two bugs
- empty pool message would show twice in a single pool when there were two empty pools (missing id meant fake machines were all grouped together)
- ensure detail page for rke1 clusters show
Co-authored-by: Richard Cox <richard.cox@suse.com>
* custom promptremove component for project and namespace remove
* Fix lin t
* Two tweaks
- Align
- createnamespace button with ns age column
- project action button with ns action button column
- Require confirmation to delete a project to match deleting a namespace
* Address PR feedback
Co-authored-by: Neill Somerville <neill.somerville@gmail.com>
Co-authored-by: Richard Cox <richard.cox@suse.com>
* Add ingressClassName field to ingress form and ingress list
Addresses https://github.com/rancher/dashboard/issues/6339
* Move ingressClass to details section on ingress detail page
Signed-off-by: Bastian Hofmann <bashofmann@gmail.com>
* provide interface to edit node name for RKE1 nodes
* prevent cluster member to go to edit because they don't have the correct permissions
* Switch to using standard done routing
- to support normal cancel/submit routing there's a number of 'done' style props/methods
- normally these are straight forward, in this case though we go back to the detail location of a different resource
- this should have worked already for cancel (doneOverride) but there was a regression
- to get this working for submit the `done()` was added which works with the components doneLocationOverride
- there's some tech debt in master atm, model doneOverride should be renamed. component doneLocationOverride should move to the model.
Addressing this though would be troublesome though
Co-authored-by: Alexandre Alves <aalves@Alexandres-MBP.lan>
Co-authored-by: Richard Cox <richard.cox@suse.com>
Bastian Hofmann