NOTE: this item needs further thinking to ensure the logic is sound and the timing between steps is reasonable.

Frequently, we see that some plugins have been updated in overlays but not built in Konflux:

$➔ ./build/scripts/checkPluginsExist.sh -v 1.9.0 
.
[1/51] [!] https://quay.io/rhdh/backstage-community-plugin-acr exists; tag 1.9.0--1.20.2 NOT FOUND!
..........................................
[43/51] [!] https://quay.io/rhdh/backstage-community-plugin-topology exists; tag 1.9.0--2.9.4 NOT FOUND!
........
Of 51 plugins:
 > 2 tags not found at quay.io
 > 49 found at quay.io for the required tag

This could mean:

• the sync-midstream.sh hasn't run yet (or is still running) so there's no new content to build in Konflux
• the Konflux job died
• the Konflux job hasn't triggered

So we could have a pipeline that runs ./build/scripts/checkPluginsExist.sh --trigger-missing to call

• .tekton/generatePipelineRunsForPlugins.sh -v 1.9.0 --regex "acr"
• .tekton/generatePipelineRunsForPlugins.sh -v 1.9.0 --regex "topology"

and then a half hour later we could have another pipeline that runs:

• ./build/ci/update-index.sh -b release-1.9

to check for the latest images and rebuild the index container, proceeding even if there are missing image refs

We may also want to revisit the logic in sync-midstream so that for no changes to workspaces, we do fewer rebuilds/yarn installs and therefore complete faster.

Since we can only have ONE pipeline script in GL, we need to implement this similar to what we have for either:

• the sync-midstream.sh for rhdh repo, that can build hub and/or operator OR bundle+FBCs, depending on what's changed in quay; OR
• the pipeline logic for calling weekly-maintenance.sh separately from sync-midstream.