Repository navigation
Promote edge to stable: RDS CA bundle pin, import keeps the latest math tick - #3063
Merged
Merged
Conversation
AWS replaced the global RDS truststore bundle (last-modified 2026-10-14 22:38 UTC), so the checksum pinned in the server and migration-runner images no longer matches and every image build fails at sha256sum. The new bundle (SHA-256 90a6f1a7..., 114 certificates) contains only Amazon RDS regional root CAs. Update the pin in both images and the migrations doc. Claude-Session: https://claude.ai/code/session_018NVGBuYCk4EZmiz4csUv9s Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The five result families keyed by a math tick (Delphi_PCAResults, Delphi_KMeansClusters, Delphi_CommentRouting, Delphi_PCAParticipantProjections, Delphi_RepresentativeComments) accumulate one copy per tick in the source: the math writer puts a new tick each run and only a full Delphi run clears the old ones. A reader never serves more than the newest tick -- it resolves Delphi_PCAConversationConfig.latest_math_tick and point-reads that one zid:tick partition -- so older ticks are history, not served state. The importer copied every tick into one result batch, so a conversation with N accumulated ticks carried N times the rows in those families. Keep only the latest tick per conversation (latest_math_tick, or the max tick present when that pointer is absent) and re-disposition older-tick rows to 'archived' through the existing accounting. Archiving, not deleting, leaves the rows in the census inventory, so source_sha256 is unchanged; only the imported and archived counts move, conservation (source = imported + archived + quarantined) still balances, and the archived rows land in archived-<family>.jsonl. Observable: those families' imported counts drop to one tick's worth and the older ticks show as archived in the dry-run report and the apply result; served results are identical (the newest tick). A fresh dry-run baseline is needed before apply because the per-family counts change. Follow-up (not here): a genuinely large single tick could still overrun result-family staging; that is a separate change. Claude-Session: https://claude.ai/code/session_018NVGBuYCk4EZmiz4csUv9s Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Delphi Coverage Report
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Pins the RDS CA bundle AWS published on 2026-10-14 (image builds were failing); the DynamoDB import keeps only the latest math tick per conversation and archives older ticks. No schema change; participant path untouched.