About adding unique index lost data - There are delete operation scenarios

[suyestyle]

Is it possible to solve this scenario?

Add two calculators when there is a delete operation.

1. The deletion times of the original table are recorded in delete1

2. The deletion times of the new table are recorded in delete2

When cut-over:

1. Count the total number of rows in the original table and record it to count1. Get the value of delete1 and calculate: res1=count1-delete1
2. Count the total number of rows in the new table and record it to count2. Get the value of delete2 and calculate: res2=count2-delete2

When res1 is equal to res2, no data will be lost.

[suyestyle]

reference only

[dm-2]

👋 @suyestyle this might be possible to do during the atomic cut-over but it would cause a delay in the cutover. We have another approach that we use at GitHub - we continuously run test migrations against real data using replicas dedicated to testing gh-ost, which gives us confidence that we're not losing data during migrations.

[suyestyle]

👋 @suyestyle this might be possible to do during the atomic cut-over but it would cause a delay in the cutover. We have another approach that we use at GitHub - we continuously run test migrations against real data using replicas dedicated to testing gh-ost, which gives us confidence that we're not losing data during migrations.

This is great.

Will this feature be added in the next stable release?