We had a site-to-site post migrations outage from sometime mid-afternoon UTC on October 27 through late morning UTC on October 30 (a little under 3 days). This is now fixed.
The actual outage was caused by some changes made to our internal network configuration. Site-to-site migrations work via API calls between sites that are configured to never leave our data center, and the changes that were made prevented these calls from going through successfully. These changes have been reversed.
This leaves nearly 3 days of site-to-site post migrations that were not performed. We are currently evaluating ways to proceed with going about performing this backfill network-wide, and will update here when we have more information to share.
- The Stack Overflow mods manually went through the 8 posts that were up for migration during this outage, and made decisions on them whether to manually migrate or not
- As for the rest of the network (where there are on average 14 migrations daily for all non-SO sites combined), we have made the decision to not put in a backfill for this. The logic needed for the backfill is actually quite complicated and unreliable (each one would need to be manually checked anyway, especially after some mods have already gone ahead and made decisions on migrate/not for all affected posts), and is not worth the investment of time to implement for a one-time event like this.
- Thus, for the approximately 50 affected posts across the network during this 3-day period, we will leave it up to mods of individual sites to act on individual affected posts and will not be performing any automated actions. Affected posts will stay closed-as-off-topic on their originating sites (and mods can manually override this and migrate them if they so choose).