Surfaced 2026-08-05 when the #917 workflow merge fired every service pipeline: spikersoft-file-movement.yml failed in create_manifest at the bao-secrets fetch —
::error::no value at 'secret/ci/backend/minio/movement' field 'secret_key' ... : []
Reproduced identically on a quiet rerun, so not the burst-rate-limit class. This was the FIRST run of the file-movement pipeline in recent history — the gap is pre-existing, not caused by today's changes. The workflow itself carries the prerequisite hint: "run openbao/provision-minio-svc-users.sh first: secret/ci/backend/minio/movement (#613)" — that provisioning apparently never happened for movement (or the flaky mc admin user add skipped the kv write; movement IS in the script's ALL_SVCS and line ~167 writes exactly this path+field, and the ci-backend policy covers secret/ci/backend/*).
Fix (needs a human with BAO_TOKEN — CI AppRole creds deliberately never live on dev machines):
then rerun backend run 19459 (POST /repos/spikerj/spikersoft-backend/actions/runs/19459/rerun) and it should go green end-to-end. Until then the file-movement service keeps running its previous image (deploy blocked, service itself unaffected).
Surfaced 2026-08-05 when the #917 workflow merge fired every service pipeline: `spikersoft-file-movement.yml` failed in `create_manifest` at the bao-secrets fetch —
```
::error::no value at 'secret/ci/backend/minio/movement' field 'secret_key' ... : []
```
Reproduced identically on a quiet rerun, so not the burst-rate-limit class. This was the FIRST run of the file-movement pipeline in recent history — the gap is pre-existing, not caused by today's changes. The workflow itself carries the prerequisite hint: *"run openbao/provision-minio-svc-users.sh first: secret/ci/backend/minio/movement (#613)"* — that provisioning apparently never happened for `movement` (or the flaky `mc admin user add` skipped the kv write; `movement` IS in the script's ALL_SVCS and line ~167 writes exactly this path+field, and the `ci-backend` policy covers `secret/ci/backend/*`).
Fix (needs a human with BAO_TOKEN — CI AppRole creds deliberately never live on dev machines):
```
export BAO_TOKEN=<admin token>
ONLY_SVCS=movement openbao/provision-minio-svc-users.sh
```
then rerun backend run 19459 (`POST /repos/spikerj/spikersoft-backend/actions/runs/19459/rerun`) and it should go green end-to-end. Until then the file-movement service keeps running its previous image (deploy blocked, service itself unaffected).
Verified 2026-08-07 — docker service inspect spikersoft-file-movement_spikersoft-file-movement
still reports Storage__AccessKey=metadata-svc on a task last updated 12:25Z today, i.e. the
merged movement-svc stack env has never rolled because the deploy still fails at the Bao fetch. openbao/provision-minio-svc-users.sh:20 and rotate-minio-ci-keys.sh:55 are both on infra
master and both honour ONLY_SVCS, so the fix really is ONLY_SVCS=movement sudo -E bash rotate-minio-ci-keys.sh with an admin token.
Status: not started (the rotation has never been run).
Closing here. Work now lives in the repo that holds the fix, so fixes #168 in a PR will
auto-close it on merge. The umbrella tracker keeps cross-repo epics only.
— Opus 5 Agent
Duplicate of umbrella #613 — same root cause, same one-command fix. Both are closed into
**spikerj/spikersoft-infrastructure#168**, which carries the richer diagnosis.
Verified 2026-08-07 — `docker service inspect spikersoft-file-movement_spikersoft-file-movement`
still reports `Storage__AccessKey=metadata-svc` on a task last updated 12:25Z today, i.e. the
merged `movement-svc` stack env has never rolled because the deploy still fails at the Bao fetch.
`openbao/provision-minio-svc-users.sh:20` and `rotate-minio-ci-keys.sh:55` are both on infra
master and both honour `ONLY_SVCS`, so the fix really is
`ONLY_SVCS=movement sudo -E bash rotate-minio-ci-keys.sh` with an admin token.
Status: not started (the rotation has never been run).
Closing here. Work now lives in the repo that holds the fix, so `fixes #168` in a PR will
auto-close it on merge. The umbrella tracker keeps cross-repo epics only.
— Opus 5 Agent
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Surfaced 2026-08-05 when the #917 workflow merge fired every service pipeline:
spikersoft-file-movement.ymlfailed increate_manifestat the bao-secrets fetch —Reproduced identically on a quiet rerun, so not the burst-rate-limit class. This was the FIRST run of the file-movement pipeline in recent history — the gap is pre-existing, not caused by today's changes. The workflow itself carries the prerequisite hint: "run openbao/provision-minio-svc-users.sh first: secret/ci/backend/minio/movement (#613)" — that provisioning apparently never happened for
movement(or the flakymc admin user addskipped the kv write;movementIS in the script's ALL_SVCS and line ~167 writes exactly this path+field, and theci-backendpolicy coverssecret/ci/backend/*).Fix (needs a human with BAO_TOKEN — CI AppRole creds deliberately never live on dev machines):
then rerun backend run 19459 (
POST /repos/spikerj/spikersoft-backend/actions/runs/19459/rerun) and it should go green end-to-end. Until then the file-movement service keeps running its previous image (deploy blocked, service itself unaffected).Duplicate of umbrella #613 — same root cause, same one-command fix. Both are closed into
spikerj/spikersoft-infrastructure#168, which carries the richer diagnosis.
Verified 2026-08-07 —
docker service inspect spikersoft-file-movement_spikersoft-file-movementstill reports
Storage__AccessKey=metadata-svcon a task last updated 12:25Z today, i.e. themerged
movement-svcstack env has never rolled because the deploy still fails at the Bao fetch.openbao/provision-minio-svc-users.sh:20androtate-minio-ci-keys.sh:55are both on inframaster and both honour
ONLY_SVCS, so the fix really isONLY_SVCS=movement sudo -E bash rotate-minio-ci-keys.shwith an admin token.Status: not started (the rotation has never been run).
Closing here. Work now lives in the repo that holds the fix, so
fixes #168in a PR willauto-close it on merge. The umbrella tracker keeps cross-repo epics only.
— Opus 5 Agent