I am unable to create indexes on Cloud Spanner tables, receiving the error, "Spanner operation failed".
Even after increasing the cluster size to from 6 to 16 nodes, I am unable to create two indexes on a ~12 million row table.
What I did:
- Created
objects
table in Spanner on 3-node cluster - Table has 10-12 columns consisting of
STRING
,INT64
, and oneARRAY<STRING>
- Primary key is two columns; a shard value (hash of
object_id
) andobject_id
- Loaded ~12 million rows
- Table had no indexes (other than the primary key) when loaded
- Loading pegged 3-node; upgraded to 6 nodes
What I tried:
- Tried to build three indexes (via DDL in console) -- received "Spanner operation failed"
- Increased Spanner node count from 6 --> 12,
- Was able to build 1 of 3 indexes (
UNIQUE
on singleSTRING
column) - Tried to build other two indexes (
UNIQUE NULL_FILTERED
on singleSTRING
columns) -- received "Spanner operation failed" - Increased Spanner nodes from 12 --> 16 (account max)
- Tried to build other two indexes (
UNIQUE NULL_FILTERED
on singleSTRING
columns) -- received "Spanner operation failed"
What else I tried (updated):
- Removed
NULL_FILTERED
clauses and tried to build other two indexes. Did not resolve, still unable to build.