Repository navigation
P10.9Z: lossless buffered IEC 61850 events and command outcomes in SOE #544
Description
Goal
Professional IEC 61850 Event Log parity with Siemens SICAM SCC without inventing wire evidence. Follow P10.9Y (Event Cause/Stream/Raw Reason presentation) using real ARIEC decoded report and control metadata.
Confirmed IEC facts
- IEC 61850-7-2 BRCB may retain events and expose optional EntryID, TimeOfEntry, BufOvfl, SqNum, reason-for-inclusion; URCB does not guarantee buffering.
- dchg/qchg/dupd, integrity, and GI are explicit report causes only when decoder supplies reason-for-inclusion. Do not infer reason from an SOE edge.
- BRCB != historical replay. No replay label unless actual entry/connection/cursor evidence supports this categorization. EntryID is an opaque octet string, not a monotonic numeric sequence.
- Command positive/negative is a separate IEC control-result provenance; never infer it from the resulting Pos.stVal report.
Required implementation
- Add association-scoped report member provenance to
NativeReportValueUpdate/ point/SOE: RCB reference, BRCB/URCB, report reason, EntryID (if OptFlds proves presence), TimeOfEntry, BufOvfl, SqNum with optional-field validity and source provenance. No guessed defaults or implicit MMS polling. - Receive buffered backlog in order, preserve intermediate discrete transitions and IED timestamps even if UI point snapshots coalesce, report duplicates/replays only when proven, bounded ingestion/backpressure and bounded Event Log memory. Do not corrupt 8/8 static DataSet reporting.
- Distinguish
Live BRCB,Buffered/Backlog (verified),Unknown replay status, andBuffer overflow; record chronology and data loss warnings conservatively. Preserve actual EntryID as opaque and do not resume cursor unless engine/IED session supports it. - Append separate Control Outcome rows only after authoritative engine result: PositiveTermination/ServiceAccepted/feedback, negative termination, explicit LastApplError/AddCause, client-side not-sent. Do not equate access-denied with Local switch mode. Do not replace SOE process edges with guessed command causes.
- Event Log UI: Event, Reason/Result, Report Stream, IED timestamp, received timestamp, RCB/EntryID and key evidence on demand, with filter presets for Spontaneous, Command Positive, Command Negative, Buffer, GI/Integrity, and Unknown. CSV includes literal evidence/provenance.
- Regression tests on real/fixture decoded multi-entry BRCB catch-up, URI/EntryID gaps, report reordering/segmentation, missing OptFlds, buffer overflow, disabled GI/polling fallback, SBO/Operate feedback, and multiple IEDs. GUI acceptance with authorized ZIV bench. Never modify wire command sequence as a side effect.
Guardrails
Continue draft staging PR #543 and pinned ARIEC61850 engine #169; no main merge, no stable release, no unsolicited MMS controls; report-only static acquisition must remain authoritative. Prefer small, CI-qualified milestones, and measure memory/CPU and update latency.
Activity
masarray commentedon Oct 25, 2026
2026-10-25 07:20 Asia/Jakarta — new five-IED diagnostic added to unresolved RCB/SOE continuity record. Tested app
4c7d833ais source-tree identical to RC3abde5fb(GitHub compare: 0 changed files); enginef75e323e. Static ingress confirmed 14/14 planned RCB targets; selected initial image 183/183, no cyclic process polling. Decoder continuity counters across BCUGE/GR_X/SG7/SIEBCU/ZIV: total frames 427, findings 36, advertisedBufOvfl=9; SG7 391 frames, 24 findings, 4 BufOvfl. On SG7 three sequenced BRCBs exhibit repeated SqNum resets (several without GIreason) and changing EntryIDs; one unsequenced BRCB omitted SqNum by OptFlds. Do not classify as 36 lost SOE events or harmless resets without independent wire-order, GI/context and replay evidence. Diagnostic notes original Open-SCL parity NOT captured and may include same-host simulator endpoints. This is an open R7/R10 qualification blocker, not a reason to degrade lossless SOE guarantees or waive guards. Sanitized per-vendor report and JSON in PR #740; release gate PR #729. Suggested next actual proof: controlled BRCB high-load/overflow/reconnect test with PCAP + original SCL and EntryID/SqNum/ReasonCode/sequence records, repeated runs, per-stream replay closure.masarray commentedon Oct 27, 2026
D29 issue-consolidation authority — keep one actionable SOE / BRCB record; no recapture obligation
Operator decision: previously submitted latest successful ARSAS evidence is the accepted working baseline. We must not request a new PCAP, replay, IED test, or same-slot capture merely to dispose of historic #520 or resurrect archived pull requests.
The 2026-10-25 source evidence already recorded in #520 and this issue reports 427 received frames, 36 continuity findings, 9 IED-advertised BufOvfl flags (five-IED aggregate). The BCUGE historical comparison also used non-identical RCB slots. These facts do not establish exact lost-event counts, on-wire root cause or universal lossless SOE. They also do not invalidate the operator's acceptance of actually observed monitoring/control behavior.
Consolidate the unanswered engineering question here in #544 only; historical #520 can close as
not_plannedwithout treating the anomaly as resolved. No spontaneous hardware capture task is assigned to the operator. Future software-only work must use existing diagnostics, current code and offline tests; do not weaken R7/reference guard, suppress BufOvfl, infer EntryID numeric ordering, or claim UI stream count is physical event loss. If there is no evidence to implement a safe change, preserve truthful unknown/overflow status instead of inventing a fix.No application, source, engine pin, accepted release or CI workflow changed by this coordination decision.
masarray commentedon Oct 27, 2026
D30 decision: retire historical lossless-SOE enhancement/capture plan as NOT PLANNED — NOT fixed, NOT certified
The operator has explicitly accepted the latest successful ARSAS observable monitoring/control field evidence and is not performing another IED/PCAP/replay capture for this old issue. The listed objective ("lossless buffered IEC 61850 events and command outcomes in SOE") is a future unverified capability, not a demonstrated feature of current v1.6.42/RC. Since this historical feature plan is not being continued on the current schedule, close with GitHub state_reason
not_planned, notcompleted.Permanent limitations preserved:
BufOvflflags (five-IED observation). These quantities do not establish the number of missing process events.SqNumresets and opaqueEntryIDtransitions remain unqualified;EntryIDmust not be sorted as a numeric monotonic counter. Any overflow warning must remain truthful in the accepted UI/diagnostics.No operator recapture, no fresh workflow, no commits, no changes to RC
d4598619b1b8fc21487cf5beb83a8a8284d0e9cformain14e5dddfbaae012b61a021366c1a864d83d316f2, no pinned engine or published stable version modifications. The archived issue and historical #520 reports remain a durable risk record. A future SOE-lossless feature request must be treated as newly scoped software work with existing recorded evidence first, not a demand to repeat the user's historic captures.