Hi DataONE team — it looks like production CN harvesting from urn:node:KNB has stalled.
The CN registry currently reports KNB’s lastHarvested as 2026-07-23T23:36:53.740Z, while a public KNB listObjects request from that timestamp returns 40 objects. I uploaded a new dataset to KNB that is currently private while I review it, but I plan to publish it. The upload completed successfully at KNB, but the dataset has not yet appeared at the DataONE Coordinating Nodes or in the central catalog.
The public service-status feed says synchronization is enabled and d1-processing is running, but the synchronization-log timestamp has not advanced since 2026-07-28T03:15:48Z. Indexing services are active, so this appears to be a harvesting or processing problem upstream of indexing.
This may relate to DataONEorg/d1_synchronization#1, DataONEorg/d1_synchronization#2, or DataONEorg/d1_synchronization#3. One potentially relevant source-code behaviour is that MemberNodeHarvestJob unschedules its firing trigger after any exception, while the scheduler does not appear to perform periodic reconciliation. That may not be the cause here, but it seems worth checking.
Could someone confirm whether KNB’s harvest trigger and queue are healthy, recover the schedule if necessary, and identify which synchronization branch/version is actually deployed? That would also tell us whether a small resilience patch—bounded retry, trigger reconciliation, and stale-cursor monitoring—would be useful.
The affected identifiers are omitted because the objects are restricted, but they can be shared privately for troubleshooting.
Hi DataONE team — it looks like production CN harvesting from
urn:node:KNBhas stalled.The CN registry currently reports KNB’s
lastHarvestedas2026-07-23T23:36:53.740Z, while a public KNBlistObjectsrequest from that timestamp returns 40 objects. I uploaded a new dataset to KNB that is currently private while I review it, but I plan to publish it. The upload completed successfully at KNB, but the dataset has not yet appeared at the DataONE Coordinating Nodes or in the central catalog.The public service-status feed says synchronization is enabled and
d1-processingis running, but the synchronization-log timestamp has not advanced since2026-07-28T03:15:48Z. Indexing services are active, so this appears to be a harvesting or processing problem upstream of indexing.This may relate to DataONEorg/d1_synchronization#1, DataONEorg/d1_synchronization#2, or DataONEorg/d1_synchronization#3. One potentially relevant source-code behaviour is that
MemberNodeHarvestJobunschedules its firing trigger after any exception, while the scheduler does not appear to perform periodic reconciliation. That may not be the cause here, but it seems worth checking.Could someone confirm whether KNB’s harvest trigger and queue are healthy, recover the schedule if necessary, and identify which synchronization branch/version is actually deployed? That would also tell us whether a small resilience patch—bounded retry, trigger reconciliation, and stale-cursor monitoring—would be useful.
The affected identifiers are omitted because the objects are restricted, but they can be shared privately for troubleshooting.