On the Sovrin TestNet, a trustee posted an upgrade transaction for 18:05 on Friday, Dec. 14. Most nodes upgraded properly, but one of them (named sovrin.sicpa.com) did not and instead got into a state where it continually wrote "in-progress" transactions to the config ledger, at a rate of about 1 every 3 seconds. This continued until a trustee wrote a transaction to the ledger restarting the indy-node service on the problem node.
Looking at the logs from the node (attached), it appears that perhaps the indy-node-control service was not running. The steward says that there were no entries in journalctl from indy-node-control, which might support this. We will debug this further with the steward, and findings will be added here as we proceed.
Meanwhile, the problem where the condition results in spamming of the ledger should be remedied, regardless of the cause of the failure to upgrade.
Attaching the config ledger from another node as well.