A couple of doc site updates from this month’s open source working group meeting:
Gray failure detection — promoted from a two-line passing mention to a proper operations guide, and now framed as a recommended production setting in suggest-only mode (paralleling how causal_read_risky is documented). The page covers the worker peer monitor, cluster-controller aggregation, the full knob inventory, trace events to watch, and a safe pre-prod → recovery → failover rollout. 7.4-only additions like the gray_failure section in status json (CC_GRAY_FAILURE_STATUS_JSON), the SS-complaint opt-ins, and the CC_ENABLE_REMOTE_LOG_ROUTER_* knob split are called out per-version. Also cross-linked from the Configuration page so it’s discoverable next to other production-recommended toggles.
What’s Next / 8.0 preview page — new Roadmap page covering work that has landed in apple/foundationdbmain but is not yet in any released 7.3.x / 7.4.x build. Top-of-page admonition is explicit that no public 8.0 build exists today. Items tracked: Stable / Cluster Health Metric (apple/foundationdb#12899), more aggressive region failover via CC_FAILOVER_DUE_TO_TPS_LIMIT_DURATION (#12849), faster GRV rejection via max_grv_queue_delay + transaction_grv_queue_rejected (#13085), VersionedMap optimization (#13028).
Feedback welcome — especially if anything looks off, or if you’d like the roadmap page to track more items.
Thanks a lot for the update. I just saw the website for the first time. It looks great, thanks a lot for the effort you have put into it.
I observed two things:
The invitation link to the Discord server has expired.
On this site, I noticed that the feature GetMappedRange is not flagged as experimental. Has it become stable? In all the docs I could find, as well as in the Java SDK (which I use), the feature is still marked as experimental. Maybe I missed something.