tag mapping with graph refresh

[cben]

Label->Tag mapping currently implemented only with save_inventory refresh.
This is last blocker for feature parity of graph refresh with old refresh (#27, https://bugzilla.redhat.com/show_bug.cgi?id=1470021).

merged

• ContainerLabelTagMapping::Mapper class, separate tag creation manageiq#16098 (merged, ignore github showing it closed)

  • refactoring, extracted Mapper class
  • Core: collect all mapped tags, batch API to create/resolve them all
    (still inefficient unbatched under the hood)
  • save_inventory_containers: use batch create/resolve
  • save_inventory_cloud: use batch create/resolve

• Use new ContainerLabelTagMapping::Mapper interface #130 (merged, ignore github showing it closed)
  Containers refresh parser: use new Mapper class, pass it to save_inventory_containers

• Use new ContainerLabelTagMapping::Mapper interface manageiq-providers-amazon#316
  Amazon refresh parser: use new Mapper class, pass it to save_inventory_cloud

• Label->tag mapper: produce InventoryCollection<Tag> manageiq#16259 Core: expose mapped tags as InventoryObjects, create tags via InventoryCollection (even for old refresh!)

• Improve tag mapping tests manageiq#16453 Test improvements

• Replace ContainerLabelTagMapping.controls_tag? with Tag.controlled_by_mapping scope manageiq#16449 reformulate ContainerLabelTagMapping.controls_tag? as a scope REVERTED

• Fix save_tags_inventory spec: assign the tag, not the catogory manageiq#16499 Fix idiotic mistake I made in save_tags_inventory spec that wasted me several days

• Replace ContainerLabelTagMapping.controls_tag? with a prefix-based scope manageiq#16501 reformulate ContainerLabelTagMapping.controls_tag? as a scope, Take 2

• Improve tag mapping specs #170 Test improvements

• Fix test errors with :container_label_tag_mapping factory manageiq-ui-classic#2788 change UI spec to use new factory

• Tag mapping in graph refresh #162 kubernetes: Use the InventoryObjects/Collections to implement mapping in graph refresh

• Tag mapping in graph refresh manageiq-providers-openshift#64 openshift: Use the InventoryObjects/Collections to implement routes, builds mapping in graph refresh, test improvements.

• Drop :tags blacklist, refresh now extracts :tags to separate collection #180: remove unecessary blacklist once routes, builds mapping is implemented.

• amazon: Use the InventoryObjects/Collections to do tag create/resolve during graph refresh
  https://bugzilla.redhat.com/show_bug.cgi?id=1506404
  Implement tag mapping in graph refresh manageiq-providers-amazon#382

not on critical path:

• 🏇 Batch queries, at least for happy path (tags that already exist). WIP.

• ✂️ Get rid of specific-tag mappings. These were never exposed in UI and add some dead code. WIP, wanted to do it first but requires massive rewrite of tests so postponed.
  refactoring, gaprindashvili/no

• ⛑️ Add way to disable mapping without deleting mapping? — I think unnecessary, fast enough now.

• 🐛 🏗️ Go through Policy
  CustomAttribute->Tag mapping should go through Policy? manageiq#15861

  • 🏗️ If policy can cancel tagging, implies moving retagging to post-refresh. Current implementation updates taggings table as a regular InventoryCollection during refresh.

• 🏇 Add unique index on Tag table! Would allow tag creation without locking.
  Migration => gaprindashvili/no

• 🐛 BZ 1508437 bug in save_tags_inventory doing nothing if collection.blank? — fixed by Do unassign tags when mapped tags list becomes empty manageiq#16370, tests in several PRs.

• 🐛 UI Add/Save/Cancel were buggy — BZ 1513507 — fixed by Fixed replace_right_cell regression in the tag mapping on OPS screen manageiq-ui-classic#2749

• 🐛 Not mapped: ContainerImage labels, Template labels, ContainerBuildPod labels (?)

  • WIP [WIP] Graph refresh: support mapped tags for images #182 images
  • WIP [WIP] Map labels to tags for templates and images manageiq-providers-openshift#73 images, templates
  • ContainerBuildPod ok, doesn't acts_as_miq_taggable

[cben]

@miq-bot add-label performance, enhancement

[simon3z]

@cben it seems we have a long list of items, any estimation on completion?

[cben]

This is long done, closing.
(except several "not on critical path" stuff, these were indeed out of scope, some now tracked elsewhere)