I’m using the Java SDK to create search indexes (FTS) on a bucket. After a bunch of indexes are created (~300 indexes), the bucket’s RAM usage is almost full (~950MiB / 1Gib) and the resident percentage reads as 0. This is causing the new inserts on the bucket to fail with this error message:
DML Error, possible causes include concurrent modification. Failed to perform UPSERT on key 92229c7c-da5e-4e85-81a6-5a02de70e2af - cause: Unable to complete action after 2 attempts
The insertions succeed after I increase the RAM quota of the bucket.
Is FTS adding extra load to my bucket’s RAM? How do I avoid the bucket from reaching it’s full quota?