On 28 July, 2025, the RPKI rsync repositories experienced degraded performance for approximately four hours during a planned data centre migration. Users may have noticed slower repository synchronization times (or, in rare cases, may have been unable to synchronize) between 09:20 and 13:24 UTC.
This incident occurred during an ongoing data centre migration project, where servers and network equipment were moved from one facility to another. Part of these servers were virtualization hosts. When those are offline for more than 60 minutes, the distributed storage system automatically initiates data repair operations to maintain data integrity and availability.
During the migration window, these repair operations generated significant network traffic to synchronize data across our infrastructure. This coincided with existing regular storage traffic, creating a network bottleneck that impacted storage performance. Since the RPKI rsync repositories rely on this storage infrastructure, users experienced performance degradation.
For subsequent server migrations later that day and the next day, the RPKI rsync repositories were temporarily moved into a memory filesystem to mitigate the risk of facing the same bottleneck again.
Based on this incident, several improvements will be implemented: