Audit all consumers that store or rely on D7 kmassets uid values

Area: solr / kmassets / migration / identity / clients Raised during: Sprint 1 (1a.8 uid identity discussion, 2026-07-01) Jira: (add when available) Priority: High — must be completed before final cutover; determines redirect and compatibility scope

Why this matters

At final cutover the kmassets index is rebuilt from D11, replacing all D7-era uid = images-{d7-nid} entries with uid = images-{d11-nid}. Any client that has stored, cached, or hard-coded old uids will break silently — returning stale data or empty results — unless identified and migrated ahead of time.

See kmassets-uid-identity-across-migration for the identity strategy and cutover steps.

Clients to audit

Client	Where to look	Risk
React KMaps app (Than)	Asset link generation; saved/bookmarked asset URLs; any place a uid or nid is embedded in a route or state	High
reindeer_x	Does it reference kmassets docs by uid when writing shadow entries? Does it read back by uid?	High
D7 Drupal content	Body fields, asset-link fields, CKEditor embeds that reference /node/{d7-nid} or asset picker selections by nid	Medium
Solr proxy	Does it rewrite or cache uid-based queries?	Low (likely pass-through)
External services	emu.shanti.virginia.edu, fox.thdl.org, cicada.shanti.virginia.edu (confirmed in terraform ALB rules as direct Solr consumers)	Medium — unknown what they store
API consumers	Any service calling api/json/{nid} or api/ajax/{nid} (out of scope for MVP per sprint doc, but still stores nids)	Medium
Saved search links / bookmarks	User-facing URLs that encode asset ids	Low (can redirect)

What to look for in each client

• Stored uid strings (e.g. images-1028396)
• Stored raw nids used to construct a uid downstream
• Queries filtered by uid:images-{nid} or id:{nid}
• Hardcoded nid ranges or nid-based routing logic
• Any database tables or persistent storage that caches asset identity

Output needed

For each client: (a) does it store old uids/nids, (b) what is the impact at cutover, (c) what is the remediation (redirect, re-index notification, data migration, or no action needed).

Suggested approach

1. Start with reindeer_x and the React app — highest risk, Than and Andres are the right people to consult.
2. Grep the D7 codebase for uid, images-, and asset-link field definitions.
3. Reach out to the external Linode-hosted services (emu, fox, cicada) to understand their consumption pattern before cutover.

Operational tracking via proxy logs

The proxy compatibility shim (see kmassets-uid-identity-across-migration) logs every request that arrives with an old-format uid. Once the shim is live, these logs become the authoritative view of which clients are still using the legacy format — use them to drive the remaining consumer update work and to know when the shim can be retired.