And we have filter on this saying vc != null

if i remove the filter this works, but with filter it is failing with this exception.

I saw the code and saw that there is no handling for ARRAY in makeProcessorInternal ? we do not expect Vcs to be Array ?

either way, if you had some multi-valued string columns before, there would be some query rewriting to do, described in these doc sections: • https://druid.apache.org/docs/latest/ingestion/schema-design.html#type-aware-schema-discovery • https://druid.apache.org/docs/latest/ingestion/schema-design.html#migrating-to-type-aware-schema-discovery

the biggest implication is this first one (the change from multi-value dimensions to arrays):

Update any queries that use multi-value dimensions (MVDs) to use UNNEST in conjunction with other functions so that no MVD behavior is being relied upon. Type-aware schema discovery generates ARRAY typed columns instead of MVDs, so queries that use any MVD features will fail.

if you aren't prepared to embark on that then please avoid

useSchemaDiscovery

for now. we're building a smoother migration path for an upcoming release, so you can wait for that

I think problem is we have a custom function which was returning a MVD.

we are on 0.21.1 [Trying to move to 26], which recent version we can fallback without breaking mvd behaviour ?

you should be OK on 26 if you aren't using the new array columns or type-aware schema discovery features; they are new & optional

if you aren't using those, & are still seeing this error, please share some additional info about your custom function and we'll try to help

I will check the details of custom function and will update. But as far as i know it take a single value and generate List<String> from it.

so the return type we changed from ``ExprType.STRING_ARRAY` to

ExpressionType._STRING_ARRAY_

assuming both are same