Hi Upspinners,
Kris Foster took me up on my suggestion to write a service that monitors our Key Server logs and alerts Upspin users if the log is tampered with, and in the process found that the incremental hashes were inconsistent.
This revealed a bug in the way the hashes were computed, where a server restart could reset the incremental hash to zero. There were three such server restarts, and so the original log was effectively divided into four incrementally-hashed sections.
A fix for the bug has been committed and deployed. To make the logs consistent again we re-hashed it all, so the hashes for all log entries after line 84 (when the server was first restarted) have now changed.
You may wish to consult the log and verify that the public key and server endpoints for your users are correct, although we have no reason to suspect they should be wrong.
Thanks to Kris for finding and reporting this issue.
Andrew