There was an error while loading. Please reload this page.
Record how to check whether the deleted storage came back The question gets asked again every few weeks and the answer depends entirely on reading the metric the right way: the plateau, not the current value, and never collStats. check_volume_reclamation.py does the reading; this says how to interpret it and which two mistakes it exists to prevent.
The reference builds now have a verified copy off AWS Taken 2 Sep 2026: all ten tarballs downloaded and checked against the md5sums published beside them in the bucket. That was the last single copy of anything this site depends on, so it comes off the uncovered list on both pages.
Add a page describing how backups actually work The off-site page covers the copies a person takes by hand. Nothing described the system: which mechanism protects which store, the policy decisions and the reasoning behind them, how to restore and what the restore looks like while it is running, and how to tell any of it is working. Also corrects the metadata dump size on the off-site page: it is ~39 MiB now, not ~51, because fs.files fell from 1,065,009 rows to 808,532 when the GridFS residue was swept.