Kibana's Dev Tool: request does not support having a body

Hi,
on 1 server RH7.6 I installed test elasticsearch & kibana 7.11.2 with ROR plugins readonlyrest-1.28.0_es7.11.2 and readonlyrest_kbn_free-1.29.0_es7.11.2.

My test command `curl -vvv -u kibana:kibana http://elk01:9200/_cat/health?v runs ok.

But when I login into kibana gui and click on Dev Tools, this command returns error:

  1. GET _cat/health (also GET _cat/indices, POST _flush/synced etc.):
    {
    “error” : {
    “root_cause” : [
    {
    “type” : “illegal_argument_exception”,
    “reason” : “request [GET /_cat/health] does not support having a body”
    }
    ],
    “type” : “illegal_argument_exception”,
    “reason” : “request [GET /_cat/health] does not support having a body”
    },
    “status” : 400
    }

  2. This command runs ok in Kibana Dev Tools:
    GET _search
    {
    “query”: {
    “match_all”: {}
    }
    }

After login into kibana gui I see these 3 messages in /var/log/messages
elk01 kibana: [17:34:50:707] [error][plugins][ReadonlyREST][cookieManager] Error: Bad hmac value
elk01 kibana: [17:34:50:746] [error][plugins][ReadonlyREST][cookieManager] Error: Bad hmac value
elk01 kibana: [17:34:50:748] [error][plugins][ReadonlyREST][cookieManager] Error: Bad hmac value

Hi peterrinka, thanks for reporting the issue.

I’m happy to inform you that it has been resolved and the fix will be included in the next release.

Regards,
Jan

2 Likes

Great news. Thank you, Jan.

We are facing the same issue.
Can we have a testbuild to test with?

We are facing the same issue. Any luck? This really makes the dev tools useless for monitoring cluster state.

This looks fixed to me in 1.31.0 (pre release)

The new stable ROR release 1.31.0 is out now :package:

I’m seeing the same error running 1.31.0 for 7.12.1.

Any ideas on how to debug?

Did you update the ES plugin as well? I cannot reproduce this at all.

Yes, I am still seeing this issue.

Kibana 7.12.1 - RoR Pro-1.31.0_es7.12.1
ES 7.12.1 - RoR 1.31.0_es7.12.1

Can you think of any other way for me to debug this issue?

It could be useful to see any logs on the ES side, and if there’s something unusual in readonlyrest.yml

@ronald.vanboven, @peterrinka do you experience the issue still?

@sscarduzio out name of @ronald.vanboven I can answer that this problem is solved on our side.

1 Like

Please let me know what would be useful to see.

ElasticSearch doesn’t output any logs for this request. Kibana only logs this line along with the request attributes.

POST /api/console/proxy?path=_cluster%2Fhealth&method=GET 400 19ms - 336.0B

I’ve verified in Chrome devtools that the Kibana front end does not send a body to the proxy endpoint.

I’ve verified that all of the ES nodes are running the correct version.

[2021-07-09T15:39:10,743][INFO ][t.b.r.b.LogPluginBuildInfoMessage$] [staging-elasticsearch-coord-only-0]Starting ReadonlyREST plugin v1.31.0 on ES v7.12.1

Have you tried cleaning up the optimize directory in Kibana after the installation of the new ROR version? When you start Kibana next time it will regenerate it.

sounds about right to me


The front end side is OK, I think it’s about the Kibana backend towards ES.

Have you tried cleaning up the optimize directory in Kibana after the installation of the new ROR version? When you start Kibana next time it will regenerate it.

This runs inside of a Docker container, and the last line of the Dockerfile is RUN rm -rf /usr/share/kibana/optimize/bundles/

In Kibana 6 we would see the optimized bundles being regenerated on startup, but I don’t see that happening in Kibana 7.

Thumbs up for the Docker choice, more reproducibility, always starts from clean slate :mechanical_arm: :robot:

Anyway, I would do rm -rf optimize/*

What about the ES side? Is this dockerised too?
Would you like to share the set of configurations (maybe in PM if you prefer)