feat: connect tournament views to APIs and recover archived stats

Remove development previews and complete hall of fame, match history, official results, and multi-kill API integration.

Fix active match pagination and team-ready controls, preserve basic scoreboards, and recover statistics when Riot archives arrive.

Use bounded in-process cache invalidation retries and add the Riot match lookup index migration with regression coverage.
This commit is contained in:
dahyeon committed 2026-11-11 02:00:35 +09:00
1 parent 20e1e6ec04
commit 0d7617e2d2
70 files changed
+3739 -1793

No files matched your search

+33
View File
@@ -117,6 +117,39 @@ The index is `{ auctionCancellationCleanupRetryAt: 1 }`, restricted to
`{ auctionCancellationCleanupPending: true }`. The worker only reads due pending
entries; it no longer scans legacy cancellation markers without a pending flag.
## Riot archive recovery
Newly saved or repaired Riot match data refreshes affected internal match
statistics at the same Competition revision. Existing archives and duplicate
imports do not trigger recovery. Recovery failures are logged and do not block
result collection or match refresh. Only communities with recovered statistics
have their caches invalidated, once per recovery batch; one failed rebuild does
not prevent other successful rebuilds from being reflected. Detail
requests also check for newly archived data when a cached game is missing raw
data. An absent archive does not trigger a rebuild or a Riot match fetch. If a
detail rebuild makes no progress, the process waits 60 seconds before retrying
that source revision and does not invalidate the community cache.
Recovery cache invalidations first try Redis with a one-second command timeout.
On failure they return immediately and schedule up to three in-process retries,
waiting 1, 5, then 30 seconds between attempts. A community has one retry budget;
additional immediate failures do not reset it, and a successful current request
cancels the pending retry. Timers are unreferenced and cleared on shutdown.
Retries are not persisted or resumed after a restart. Detail requests detect
newly archived raw data even behind a stale cache, and other response caches
expire after five minutes. A detail response whose immediate invalidation fails
is returned without being cached. No Mongo invalidation collection or periodic
queue scan is used.
Before deploying this recovery path, add its Riot ID lookup index. The migration
defaults to dry-run and leaves other indexes and game records unchanged.
```bash
pnpm --filter api stats:indexes -- --env staging --dry-run
pnpm --filter api stats:indexes -- --env staging --apply
# Use --env prod for production.
```
## Deployment
When you're ready to deploy your NestJS application to production, there are some key steps you can take to ensure it runs as efficiently as possible. Check out the [deployment documentation](https://docs.nestjs.com/deployment) for more information.