We have not observed any further errors and all previously impacted sites are fully operational.
Equinix has put in place a fix and shared that the outage was caused by an erroneous prefix list update based on an incomplete prefix list returned by RADb (https://www.radb.net/
) during an outage.
Render will continue to explore redundancy options to mitigate the failure mode that occurred here in the future.
Customers who implemented a temporary A record change per the mitigation suggested earlier in the incident must revert that change as soon as possible and point back to the old IP.