I deployed a 2-region, 3-DC FDB cluster and configure it with:
{
"regions": [
{
"datacenters": [
{
"id": "dc1",
"priority": 2
},
{
"id": "dc2",
"priority": 0,
"satellite": 1
}
],
"satellite_redundancy_mode": "one_satellite_double",
"satellite_logs": 24
},
{
"datacenters": [
{
"id": "dc3",
"priority": -1
}
]
}
]
}
and then set the usable_regions to
1.
I loaded my data into dc1
. After loading completes, I set usable_regions
to 2 so that the data is copied over dc3
. After several hours, I checked the status using fdbcli
and see this:
Data:
Replication health - Healthy (Rebalancing)
Moving data - 0.000 GB
Sum of key-value sizes - 2.176 TB
Disk space used - 16.832 TB
It looks like the data movement completed. However, when checking status json, the data version difference is still -495671
(I expect it to be 0 since two data centers should have exactly the same data). My questions are:
- If the data movement completed, why the data version difference is not zero? If data is still moving, how could I know when the movement completes?
- Sometimes I see the data version difference is positive, sometimes it is negative. What does that mean? How the data version difference is computed then?