aiohttp 3.14.5 buffers 256 KiB and pauses reading above 512 KiB; name
the aiohttp versions actually checked instead of a range; fix line
references in the idle-connection evidence; the object-storage spec now
exempts the connect timeout from the single call timeout.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
aiohttp 3.13-3.14.5 re-arms sock_read in resume_reading() after a
streamed response has already returned the connection to the pool; the
timer fires while the connection idles and the next call on it fails in
1 ms with ReadTimeoutError (503 on a healthy S3). The call and per-chunk
asyncio.timeout stay the only S3 timeout. Adds driver.py s3-idle and
documents the timeouts in the rules.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Keep the committed mark, then delete the object and the row under the
same row lock, so an older replica without the mark check that waits on
the row finds it gone instead of promoting it to ready after the object
delete. A mixed-version stand is no worse than before the mark; document
the upgrade order (replicas first, cleaner last) and fix stale wording.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Mark a stale pending row in its own committed transaction before the
cleaner deletes the object, and let mark_ready promote only unmarked
rows, so a cleaner rollback after the object delete can no longer leave
a ready attachment without its object. Add fault checks for an unknown
COMMIT outcome, an upload outliving the grace period and a cleaner
failure after the object delete; fix the cleaner entrypoint in rules.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- upload: delete the stored object when ready fails before COMMIT or the
pending row is gone (503 instead of 500); unknown COMMIT outcome keeps it
- cleaner service (python -m src.cleaner): stale pending rows older than
grace (900 s, must exceed 17 s live-upload bound) under FOR UPDATE SKIP
LOCKED, object before row, own S3 user with read/delete only
- shared session_scope for API and cleaner; early_stop process name
- partial index on pending attachments, CREATE INDEX CONCURRENTLY
- s3-bootstrap: cleaner user, policy and checks; make env_setup appends
missing template variables to an existing .env
- failure checks: cleaner on the failure stand, id-list hook actions,
five new and three updated tests; driver upload-consistency
- sync upload-consistency deltas into main specs (storage-consistency)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- keep one upload body in memory: BytesIO + getvalue() without copy,
fixed glibc mmap threshold in the image (no realloc copy on growth)
- stop PNG inflate at expected size + 1 byte
- reject filename that is not valid UTF-8 after percent-decoding (400)
- failure stand: env limit variables do not change attachment limits;
history with attachment_ids while S3 is stopped
- drop dead constants; restate "old image on new schema" scenario
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- upload, metadata and download endpoints for PNG/JPEG attachments:
limits before reading and decompression, type by content,
structural integrity check without decoding pixels, safe filename
- write order: pending row with object key -> PUT to S3 -> ready;
per-step DB sessions so no pool connection is held during transfers
- attachment_ids in messages (validation, idempotency hash, history);
whitespace-only text allowed with an attachment (teacher, 2026-10-20)
- migration: attachments and message_attachments tables, text length
CHECK relaxed with NOT VALID
- student tests, fault checks and driver commands for attachments;
rollout profile starts from e1161f7 and checks the commit first
- sync attachments deltas into main specs; rules, AGENTS.md, PLAN.md
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- s3 healthcheck and fault hook probe wait for the volume server to
register with master (first write after restart no longer waits ~5 s)
- s3-bootstrap mounts /data as tmpfs: no anonymous volume per container
- S3 client: NoSuchBucket and AccessDenied are 503 with their own error
logs, not a missing object or rejected credentials; delete ignores only
a missing key
- test_isolated verifies prefix removal (weed shell exits 0 on errors)
- s3_check derives ports from listening sockets in the s3 container
- readiness without S3 is also checked with s3 paused
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- s3 and s3-bootstrap services: SeaweedFS 4.48 pinned by digest,
internals on loopback, signing key, private bucket, API user policy,
privacy checks on every bootstrap
- [s3] settings and aiobotocore client in src/infra/s3: deadline per
call, no retries, 503 on outage or rejected credentials
- make start_s3, stop_s3, s3_check; run_dev, test_isolated, wipe_data
and restart hook include S3
- failure_checks with own S3 and readiness check without it; bench
generates S3 secrets and filters them out of the passport
- rules, AGENTS.md, skill; sync health delta; KNOWN_ISSUES: rollout_from
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- carry the lab 3 solution over from mtusi2026lab3-safronov
(solution/bvt2301-safronov, ce89eba): src, migrations, student-tests,
bench, openspec, .claude, infra/haproxy and infra/valkey, own scripts,
Dockerfile, compose, pyproject and uv lock
- merge Makefile, .gitignore and .dockerignore (the lab 3 versions only
add lines to the skeleton, which is the same in lab 4) and
.env.example (lab 4 skeleton names listed in the header)
- rename the compose project to mtusi-lab4 and the image to :lab4 so
the stand does not reuse the lab 3 containers and volumes
- update AGENTS.md for lab 4: current lab, origin commit, contract and
test files of the new skeleton
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>