Introduce process-local and Redis-backed global rate limiting for the
shared LLM client, plus a dedicated generation consumer prefetch knob.
- Add request (RPM) and token (TPM) rate limits with burst controls,
process or global scope, and fail-closed/process-degraded failover
- Wire the LLM client through NewWithRedis so limits can be shared
across replicas via a Redis token bucket
- Add generation_consumer_prefetch to tune RabbitMQ generation
consumers independently of the default prefetch
- Bump generation worker_concurrency to 24 and promote golang.org/x/time
to a direct dependency
- Drop t.Parallel() from a few middleware/bootstrap tests
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Ship the Ops backstage 对象存储 page (browse / upload / move / delete /
folder ops) backed by a new object-storage service + handler, with the
shared storage client extended with List/Stat/Copy/Usage/DownloadURL so
both MinIO and Aliyun providers cover the new surface. Add
ForcePathStyle to the object-storage config and treat r2/s3/custom_s3
as external providers so Cloudflare R2 and other S3-compatibles can
share the MinIO client path without spinning up the bundled MinIO.
Also add CSV import/export + template download to the site-domain
mappings page, raise gin MaxMultipartMemory to 8 MiB for the new
multipart endpoints, register Ops swagger routes, and extend the
descriptions-parity test to cover the ops router.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
root