Hi,
4.4.1 Commit SHA 57474ae
21 Visualize: Cannot read property 'responses' of undefined
TypeError: Cannot read property 'responses' of undefined
at Object.getResponses (http://10.72.58.246/bundles/kibana.bundle.js?v=9693:80248:21)
at http://10.72.58.246/bundles/kibana.bundle.js?v=9693:78670:26
at processQueue (http://10.72.58.246/bundles/commons.bundle.js?v=9693:42357:29)
at http://10.72.58.246/bundles/commons.bundle.js?v=9693:42373:28
at Scope.$eval (http://10.72.58.246/bundles/commons.bundle.js?v=9693:43601:29)
at Scope.$digest (http://10.72.58.246/bundles/commons.bundle.js?v=9693:43412:32)
at Scope.$apply (http://10.72.58.246/bundles/commons.bundle.js?v=9693:43709:25)
at done (http://10.72.58.246/bundles/commons.bundle.js?v=9693:38158:48)
at completeRequest (http://10.72.58.246/bundles/commons.bundle.js?v=9693:38356:8)
at XMLHttpRequest.requestLoaded (http://10.72.58.246/bundles/commons.bundle.js?v=9693:38297:10)
Getting this when executing a dashboard with 22 visualisations against an index with 17m documents. However all is fine when running it against a different index with 21m documents.
I haven't been able to replicate this. Best bet is here:
The only clue I have is bad data. Had this problem when one of indexes was trashed due to buffer overflow durring indexing. I might still try to replicate it, hopefully.
I am getting same/similar error. Observing the network call on "/_msearch" call, it looks like it was transferring a large dataset (~480 MB in my case) and it errors out.
There must be scenarios where the elasticsearch client will respond with a good status, but not have a body.
The core issue here is that we aren't validating the validity of the response we get back around here. Maybe this should happen in each of the strategies, or at a higher level, or maybe it's a bug in the client.
I agree though, we need a reliable way to reproduce this.
Closing due to lack of activity in response to "needs details" label. Please re-open if this problem is reproducible with the latest version of Kibana.
I just ran into this (ES 2.4.2) as well. The error seems to occur when one of the requests times out, because it takes a while, and there's no response body when I look at the last msearch request that kibana made in devtools.
@tbragin I am getting this error in version 5.2.1 on Kibana. It was happening to some of the indices values when I try to load the visualization in Kibana dashboard that too when I change the time interval for a week, year and month for the indices.

I inceased my elasticsearch timeout interval from 3000 to 9000ms. I have 1 shards and 2 replica indices.
We are getting this error in Kibana/Elasticsearch 5.2.1 and 5.5. The error occurred when running aggregation that has a very large response payload. We observed in Chrome's dev tool that the issue occurred after ~900MB of the response transferred. The same aggregation runs without issue when run from Safari. Increasing Chrome's "max_old_space_size" to 4GB didn't help fix the issue.
+1
Got the same error on Kibana 4.5.4. Only some dashboards/visualizations, and only on one specific index.

Above picture is from Google Chrome browser - Version 59.0.3071.115 (Official Build) (64-bit)
Safari (Version 10.1.2 (11603.3.8)) works well.
Most helpful comment
I just ran into this (ES 2.4.2) as well. The error seems to occur when one of the requests times out, because it takes a while, and there's no response body when I look at the last msearch request that kibana made in devtools.