tapir

test: two more tests derive their port from the PID (55000 + pid%1000) and can collide like the store test did

opened 2026-10-10 09:02:06 +00:00 by mathias

Upload idle bound does not enforce a minimum rate: a 1-byte-per-59-s drip still holds a receive slot risk/medium type/fix

#64

opened 2026-10-04 20:58:04 +00:00 by mathias

Uploads: choose the transcription language (Swedish / other / auto-detect), route to the right engine

#61

opened 2026-10-04 07:08:09 +00:00 by mathias

v0.32.0 - audio upload follow-ups

4

Per-recording choice: keep this recording on local models only (prerequisite for allowing Berget as reconciler)

#60

opened 2026-10-04 06:46:44 +00:00 by mathias

v0.32.0 - audio upload follow-ups

serve exits when the OIDC issuer's discovery returns 503 at start-up

#59

by mathias was closed 2026-10-04 20:52:59 +00:00

v0.32.0 - audio upload follow-ups

1

Shutdown: log an undrained Shutdown and let a second signal kill the process risk/low size/s type/fix

#57

opened 2026-10-03 19:13:13 +00:00 by mathias

v0.32.0 - audio upload follow-ups

Slow-drip upload body holds a receive slot without a time bound risk/medium size/s type/fix

#56

by mathias was closed 2026-10-05 04:59:29 +00:00

v0.32.0 - audio upload follow-ups

4

Sprint: audio upload follow-ups (v0.31.1 hotfix, then v0.32.0) risk/medium type/chore

#54

opened 2026-10-03 18:50:39 +00:00 by mathias

v0.31.1 - security and robustness

1

LLM client reads an unbounded response body before checking the status risk/low size/xs type/fix

#53

by mathias was closed 2026-10-03 22:40:55 +00:00

v0.32.0 - audio upload follow-ups

1

Say when a recording's audio went to the cloud: label the serving endpoint, and document that an empty transcript also triggers the fallback risk/low size/xs type/feat

#52

by mathias was closed 2026-10-04 06:46:56 +00:00

v0.32.0 - audio upload follow-ups

2

Progress tracker: a hung database save blocks reporting, and the heartbeat uses a different clock from the page risk/medium size/s type/fix

#51

by mathias was closed 2026-10-03 22:40:55 +00:00

v0.32.0 - audio upload follow-ups

1

SIGTERM is not handled: every deploy kills the process without draining or failing the running row (interim fix; the full drain is recorded as a limitation) risk/medium size/xs type/fix

#50

by mathias was closed 2026-10-03 22:40:34 +00:00

v0.31.1 - security and robustness

2

ffprobe/ffmpeg resource limits: a 900-stream file costs about 570 MB against the 512 Mi pod limit; no probe deadline; unbounded probe output risk/high size/s type/fix

#49

by mathias was closed 2026-10-03 19:22:06 +00:00

v0.31.1 - security and robustness

1

Security: refuse cross-origin POSTs; POST /account/delete is forgeable from a sibling *.d-ma.be origin risk/high size/xs type/fix

#48

by mathias was closed 2026-10-03 19:22:06 +00:00

v0.31.1 - security and robustness

1

Audio upload: cap concurrent uploads in flight, and give ffprobe its own deadline (found reviewing the production enablement, infra#520)

#47

by mathias was closed 2026-10-03 19:22:07 +00:00

v0.31.1 - security and robustness

2

Audio upload hardening follow-ups the first adversarial review could not verify or did not fix risk/medium

#46

opened 2026-10-02 14:53:02 +00:00 by mathias

v0.32.0 - audio upload follow-ups

2

DECISION NEEDED: evaluate berget.ai models as the reconciler? It means sending private transcript text to a cloud API risk/high type/decision

#45

opened 2026-10-02 14:23:40 +00:00 by mathias

v0.32.0 - audio upload follow-ups

1

Reconcile: store every adjudicated spot, and keep saved segments in step with the merged text

#44

by mathias was closed 2026-10-03 14:13:58 +00:00

2

Upload worker logs nothing until the job ends: no start, phase, engine or adjudication lines

#43

by mathias was closed 2026-10-03 14:13:56 +00:00

2

Upload page: show progress, so the user can tell running from hung or crashed

#42

by mathias was closed 2026-10-03 14:13:37 +00:00

2