@MichelDiz this is already available through the Alter API. You can drop indexes and then, add them back again and Dgraph will build the indexes again.

That is not true. When you drop the predicate it will DELETE the tablet. The objective here is to have a non-destructive rebuild.

Please, close an issue with proves. I couldn't find any "drop indexes only" or "rebuild indexes" in the Schema API.

There is only:
dropAll
dropData
dropAttr
dropNs(namespace)

PS. If you don't get the objective, ask to clarify. I can improve the text to be more clear if the case.

How will that benefit? Internally the implementation will drop and recreate the indexes?

I don't believe there are any drops(for data). Dgraph has always done this, rebuilding the index on the fly with each new Schema change. It was after I requested for this to stop that we didn't do it anymore. I remember Martin doing that. Since then I noticed that users were complaining about indexing. So I suspected it was this change.

We didn't drop the data. We just deleted the index and then we recreate it.

Here the issue #2828

That was one of many changes the devs did back then. In the reindexing and schema code in general.

I still am not sure what is expected here. Maybe elaborate more. Currently we have enough of options for the user to delete attributes, any indexes (one or more) and recreate them.

Rebuild ≠ delete attributes

any indexes (one or more) and recreate them

How? I believe we are not talking about the same thing.

And there is a way to rebuild the index, but it's not obvious. And it's not professional either.
The way is, remove the index, wait and put it back again. This is not obvious.
Let's say the user really needs to rebuild indexing. We haven't documented anything about that. We do not recommend anything about this anywhere. They learn by trial and error. So it's better to have an explicit function than a "hidden way".

In Neo4J you can do

SHOW INDEXES YIELD name RETURN 'DROP INDEX '   name  ';'

And this will drop only the index table, not the data or the schema. They also have other functions like "verify".

Rebuild indexes is a common thing in any Database. Not sure why you are so confused with the concept.
You can throw on google "rebuild index" and you gonna see thousands of examples in SQL and others.

The point is that, if this option exists in the market of databases, why we don't? Is there any specific reason?
Rebuilding indexes are considered a maintenance function due fragmentation.

quote:

The script below allows you to rebuild indexes for all databases and all tables within a database. This could be further tweaked
to handle only indexes that need maintenance based on fragmentation levels as well as then doing either an index reorg or an index rebuild.

The script uses two cursors one for the databases and another for the tables within the database.

Rebuild All Indexes Script for SQL Server 2005 and Later

https://www.mssqltips.com/sqlservertip/1367/sql-server-script-to-rebuild-all-indexes-for-all-tables-and-all-databases/

ALTER INDEX oldindex REBUILD PARAMETERS('tablespace=TBS_3');

https://docs.oracle.com/database/121/SPATL/alter-index-rebuild.htm#SPATL1017

Optimize index maintenance to improve query performance and reduce resource consumption

https://learn.microsoft.com/en-us/sql/relational-databases/indexes/reorganize-and-rebuild-indexes?view=sql-server-ver16

I think I see your point now. You are right that we don't have an explicit API for rebuilding indexes.

This issue has been stale for 60 days and will be closed automatically in 7 days. Comment to keep it open.