Skip to content

Server products, admission and recovery

Curated source-specific evidence, not a new run or qualification of current source. Dates, measurements, failure outcomes and original source identifiers retain their recorded meaning. Old tags/assets have been retired and original commits may no longer resolve after the privacy rewrite. Where identifiers were sanitized earlier, the original hardware/run association cannot be independently recovered from the public derivative. Current acceptance lives in the roadmap.

The operator accepted Server Phase 8 and narrow Phase 9 consumption on 2026-11-09 at 744b840a5c9b3c15967d84d45e777b686779c8da. Main run 42536097284 passed selected source checks; release run 42536106408 resolved required exact-source proofs and passed both installed architectures. The retired q1 release failed a configuration-reader lock race; q2 used persisted authenticated API reads. These statements preserve the original source association, not qualification of the current branch.

Selected supporting records

Record Specific purpose
Runtime Worker containment, resource admission, durable publication and recovery contracts
Tasks Checkpoint, mounted-fault and installed outcomes, including failed attempts and scope corrections
WAL Real WAL capacity exhaustion/recovery; ACK and storage boundaries
Imagery Independent native imagery/reference comparisons, source digests and explicit exclusions
Fairness One-worker foreground/backfill starvation regression and bounded fair admission
Browser Bounded product discovery, durable import and browser retention; fixture limits
Processing repair ProtectProc/non-dumpability observation-handle defect, before/after evidence and failed qualification

JSON outcomes are unchanged. Their then-pending flags describe those executions, not today's acceptance. The routine enrollment fixture still expecting obsolete "unavailable" text and nonblocking synthetic handoff EAGAIN before the WAL assertion were fixture failures, not successful WAL tests; those redundant handoff receipts are omitted. The substantive tasks/WAL records retain results.

Limits

Phase 6 sustained performance acceptance remains open. The 2026-11-02 sanitized scheduler hit the five-second archive-I/O watchdog on an 822-byte commit; ACKs stopped safely. A targeted rerun did not explain the stall. Later green CI is not a diagnosis. Preserve the failure deadline and assertions.

Memory64 >4-GiB, wider devices/Safari/iPad, long resources/soak and integrated multi-server behavior remain unqualified where no matching evidence exists. Emulated ARM64 installed tests are not physical Pi performance. Synchronization and SIGKILL recovery do not prove arbitrary hardware power-loss behavior. Issues #5 (hidden-main radar), #6 (Full Disk cold load) and #7 (mount-table trigger/recovery, deferred unless recurrent) remain independent follow-ups. See the Server roadmap.