Unverified50% confidenceCautionExact time
InnoDB's old versions do not disappear automatically; the purge thread must asynchronously clean them up after confirming that no earlier active transaction references them; long transactions hold back purge, causing undo space to bloat.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedInnoDB's write path is log-first: any modification is first written to the redo log, then dirty pages are flushed asynchronously; after a crash, recovery is done by replaying redo, and rollback and multi-version reads are achieved through undo.61% similarUnverifiedMethods for dealing with stale embeddings include attaching version timestamps and validity flags to chunks for metadata filtering, introducing recency-based re-ranking after retrieval, and hard isolation of deprecated versions.60% similarUnverifiedThe claude project purge command has been renamed to claude purge; the old name still works but will print a notice.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1026223API
curl https://kongchang.com/api/v1/knowledge/claims/1026223MCP
get_claim(id=1026223)