Cannot apply settings to all indices

[ferronrsmith]

Elasticsearch version (bin/elasticsearch --version):
7.0.0-beta1
OS version (uname -a if on a Unix-like system):
docker
Description of the problem including expected versus actual behavior:

Cannot apply settings to all indices

Steps to reproduce:

curl -X PUT \
  http://localhost:9200/_settings \
  -H 'Content-Type: application/json' \
  -d '{
    "index.blocks.read_only_allow_delete": null
}'

Returns the error below :

Provide logs (if relevant):

{
    "error": {
        "root_cause": [
            {
                "type": "index_not_found_exception",
                "reason": "no such index [null]",
                "resource.type": "index_expression",
                "resource.id": "_all"
            }
        ],
        "type": "index_not_found_exception",
        "reason": "no such index [null]",
        "resource.type": "index_expression",
        "resource.id": "_all"
    },
    "status": 404
}

Refer to doc : https://www.elastic.co/guide/en/elasticsearch/reference/7.0/indices-update-settings.html

The REST endpoint is /_settings (to update all indices) or {index}/_settings to update one (or more) indices settings. The body of the request includes the updated settings, for example:

[elasticmachine]

Pinging @elastic/es-core-features

[ebadyano]

I am able to reproduce the behaviour when elasticsearch instance doesn't have any indexes. If I have at least one index I cannot reproduce this error.

We might want to have a slight more clear error message than no such index [null] for such cases (when there are no indexes)?

@ferronrsmith Just to confirm, are you able reproduce this error when you have more than 0 indexes?

[gwbrown]

I also see the same behavior - this call works if any indices exist, but fails with index_not_found_exception if there are none.

This also occurs on the latest commit of 6.7, but I haven't checked any earlier than that.

[elasticmachine]

Pinging @elastic/es-core-infra

[ferronrsmith]

@ebadyano confirmed; it only happens when cluster has no indices.

[hub-cap]

It looks like this is where the resolution is happening and failing. I think we should check to see if any indices exist and if not, throw a IndexNotFoundException(String message, String name), and put some information in the message regarding no indices existing.

This would be a great starter issue for someone who wanted to contribute.