I have a couchbase server (bak01) which backs up up a remote cluster cluster (cls01), I also have a cluster I'm testing restores to once a week (rst01).
Bak01 does a full backup of cls01 once a week and hourly incrementals.
I can query the full+increments and get a complete item count (lets say its 500). When I restore the backup to the restore cluster (rst01), the item count is different (backup has more items than restore - let's say 300). Strangely though, rst01 vs cls01 have the same number of items.
So three questions:
- Why is the item count different? I would guess it's because deletes are included as an item; but I'd want that clarified ideally.
- Is there a better metric I can use for restore validation? Lots of metrics are available from the backups, but less from live services; I need to compare them somehow, so consider what's available on a live system vs backup metric.
- If 2 is a no, is there a better method of getting an accurate representation of the item count? I can change the query to remove something specific if appropriate (I took the tombstones off the total and still got the same count).
Thanks
