Multiple join-split failures, global rescan enforced
Currently Horizen has a challenging issue to be solved: orphan blocks associated to transaction limitations. This issue is been addressed by the team and a solution may become satisfactory by the change to sampling.
Until the improvement, the expected orphan blocks will continue to cause join-slit erros during the challenge proof. Since we noticed too many nodes with orphan blocks we decided enforced a global -rescan to prevent/mitigate reward loss.
This is a technology issue that cannot be predicted or prevented, only fixed when it happens. Hence the loss of a single reward due this issue is excluded at our SLA, as is an exclusion cause even for compensations by its own project.