Elevated upload latency in Frankfurt
Resolved after 47 minutes. One storage node was rebuilt after a disk controller fault; writes were served by the remaining two nodes throughout and no objects were lost. Reads were unaffected.
Updated continuously from internal health checks.
Resolved after 47 minutes. One storage node was rebuilt after a disk controller fault; writes were served by the remaining two nodes throughout and no objects were lost. Reads were unaffected.
Resolved after 22 minutes. A configuration push shortened the negative cache window; objects written during the window returned 404 from the delivery endpoint for up to a minute after upload. The API itself was correct throughout.
Planned work is announced here at least five days ahead and is scheduled between 02:00 and 05:00 UTC. Nothing is currently scheduled.