Long _source_excludes/_source_includes passed in query parameter, HTTP 400 'too long http line exception'

[plexaikm]

Using search with list of long named parameters in es.search _source_excludes/_source_includes fails with HTTP 400 'too long http line exception'

Elasticsearch 8.14.1 and python Elasticsearch client 8.14.0

es.search(
...
_source_includes=[... list of fields ...],
_source_excludes=[... list of fields ...],
...
)

POST .../_search:
the _source_excludes/_source_includes are passed as query parameters bloating the http line,
can be passed in request body _source.excludes/_source.includes to avoid this issue

GET _mget:
same solution can be applied

Same issue with GET .../_doc/... but there no easy way around

[pquentin]

Hello, and thanks for your question. Unfortunately, there's nothing I can do with the client, this is the default value in Netty. If you're hosting Elasticsearch yourself, you can tune http.max_initial_line_length wich default to 4096.

Otherwise, you can use the mget API or try to reduce the list of fields, possibly by using regexes.

[plexaikm]

Hi,
Thank you for the reply and suggestion to alter parameter, yes it is self hosted so I can try it

For search/mget the client can put the lists of includes/excludes in body and not in query parameters

I'm trying a local workaround using es.search(body=es_body) which is deprecated

es_body = {
"query": ...
"_source": {
"includes": [... list of fields ...],
"excludes": [... list of fields ...]
}
...
}

[pquentin]

This is no longer deprecated, what version are you using? You can also pass individual parameters like source and query.

That said I would caution against using search, it is more expensive than get or mget and requires a refresh for read after write operations.