Repository navigation
Description
Import summary, 2026-W45
Needs attention: Neurobagel verification is alarming; 1 part(s) of this report failed.
Covers 2026-11-02T03:00:20.120Z to 2026-11-09T03:00:20.120Z (UTC). Generated 2026-11-09T03:00:20.120Z.
cc @nemarAdmin
Imports
| Imported this week | 2 |
| Managed OpenNeuro mirrors in total | 620 |
| In scope on OpenNeuro | 634 |
| Mirrors no longer in scope upstream | 8 |
Coverage
Verdict: HEALTHY. 0 in-scope dataset(s) outstanding, below the alarm threshold of 5. Last dispatch: 45 hours ago.
Datasets outstanding (nothing is working on them): 0
Upstream freshness
Examined 614 of 614 managed OpenNeuro mirrors (6 withdrawn and 0 other rows left out); 525 of them were compared with upstream.
| state | mirrors |
|---|---|
| At upstream's latest snapshot | 457 |
| Behind upstream, new tag on the GitHub mirror | 25 |
| Behind upstream, new tag not on the GitHub mirror yet | 0 |
| Behind upstream, GitHub mirror not checked | 38 |
| Upstream and NEMAR disagree | 5 |
| Tag unknown (not compared) | 42 |
| Tag never examined | 0 |
| No NEMAR version yet | 1 |
| Import running | 1 |
| Parked | 42 |
| Could not be compared (OpenNeuro unavailable) | 3 |
Stale in total: 63. Tag unknown and probably stale by date: 2 (display only, not counted as stale).
A stale mirror is reported, not an alarm: being behind upstream is an expected state, not a fault. Mirrors with an unknown tag, never examined, with no version yet, with an import running or parked, and those whose source id does not match were not compared.
How the three "Behind upstream" rows divide depends on the 25 GitHub checks one weekly pass makes; it is not a fleet-wide proportion.
Names and reasons: nemar admin upstream-status.
Upstream re-pulls
Open switch (AUTO_REPULL_ENABLED): on. Merge switch (AUTO_REPULL_MERGE_ENABLED): off. Daily open cap: 2 (the default). Daily merge cap: 5 (the default).
The scheduler's daily tick last ran 2026-11-09 03:00:26 UTC and concluded ran. The advance tick last ran 2026-11-09 03:00:09 UTC.
| Dispatches sent this week | 0 |
| Opened a pull request | 0 |
| Finished after a merge | 0 |
| Refused by the run | 0 |
| Failed in the run | 0 |
| Not reported in time | 0 |
| Not accepted by GitHub | 0 |
| Still running | 0 |
| Open automatic re-pulls now | 0 |
| Of those, held | 0 |
| Claims to merge held over an hour | 0 |
| Re-pulls that ended failed this week | 0 |
| Merged by the Worker | 0 |
| Merged by an administrator | 1 |
| Merged outside NEMAR's tools | 1 |
| Merged, claim not in the audit log | 0 |
A dispatch is counted in the week it was sent, in the state it has when this report is made. A switch that is off, a re-pull waiting on a gate (the identifier screen, the review, BIDS validation) and a run's refusal are expected and are not an alarm. What asks for attention is a scheduler that stopped while a switch is on, a scheduler that waits for the two first re-pulls to be run by hand while a switch is on, a dispatch nobody answered or that GitHub did not accept, a claim to merge held for an hour, a re-pull that ended failed, a re-pull that used all its resumes or all its finish runs, and a cap that cannot be read; every other hold is counted in its table and does not ask.
Names and reasons: nemar admin repull auto.
Dispatch liveness
AUTO_IMPORT_ENABLED |
true |
| Last dispatch | 45 hours ago |
| Dispatches landing | yes |
Open failures by cause
| cause | datasets |
|---|---|
data-unavailable |
1 |
needs-triage |
65 |
Total: 66.
What the daily sweeps did this week
Tracking issues closed on recovery: 0. Relabelled after a cause change: 0.
Neurobagel federation
Latest daily run (2026-11-08T03:00:20.318Z): ALARM.
| check | verdict | what it found |
|---|---|---|
| store | alarm | 1 eligible dataset(s) have been missing from the store for more than 48 hours. |
| node | healthy | The node serves 907 valid, protected record(s), all for eligible datasets. |
| registration | unchecked | NEUROBAGEL_FEDERATION_URL is not set, so registration is not checked. |
| drift | alarm | Upstream has moved from the pins: query tool v0.17.0 -> v0.17.1; node API v0.11.0 -> v0.12.0; vocabulary config_metadata/config_namespace_map.json changed. |
Daily runs in this window: 7 of 7 expected (healthy 0, alarm 7, unknown 0, unchecked 0; the sweep itself failed on 0).
Checks that did not run: registration 7 day(s).
Findings that need a person, over the window and what stands: 0. A count only: the audit log holds the detail.
Blocklisted datasets
| dataset | reason | parked |
|---|---|---|
| on008940 | upstream_403_after_window | 26 days |
Parts of this report that failed
Listed rather than omitted, so a gap is never mistaken for a zero.
upstream: 3 mirror(s) could not be compared with OpenNeuro
Posted weekly by the import summary job whether or not anything is wrong (nemarOrg/nemar-cli#1312), because a report that only appears when something breaks cannot tell a healthy week from a broken reporter. Last week's summary is closed automatically when this one is filed; its content stays readable.
Activity