Kibana's Upload CSV feature isn't intended for gigantic import jobs, so
I originally set a sane default of 1GB. Some users exprssed a desire to
import slightly larger files, they should be able to import something
that's 1.1GB without being blocked by an arbitrary limit. So I've made
the limit configurable via kibana.yml.
This change includes a few pieces:
* Added optional `kibana.addDataMaxBytes` key to kibana.yml with 1GB
default
* Set upload data route payload limit based on new config value
* Updated help text in UI to use the dynamic config value on the parse
csv page
* Updated parse csv page to check file size and fail early if the
selected file is too big
Resolves: https://github.com/elastic/kibana/issues/7671
The current version in docs is always trailing the current version of
the project, since the current version of docs is for the latest release
of this version line rather than the version that is currently in
development.
A new server-side configuration, elasticsearch.customHeaders, allows
people to configure any number of custom headers that will get sent
along to all requests to Elasticsearch that are made via the proxy or
exposed client.
This allows for advanced architectures that do things such as dynamic
routing based on install-specific headers.
When following the description from the docs to set the elasticsearch url I got a warning that the key was deprecated. This updates the docs to use the non-deprecated key. I haven't checked if this needs to be updated in other parts of the documentation too.
When polling the NodeJS server for operations data every 10 seconds, the
data become coarse-looking when viewing it at fine-grain detail. If you
aggregate the data into a date histogram and the bucket size is 10
seconds, then there'll be missing data in every other bucket. That is,
unless the time bounds of the buckets line up PERFECTLY with the time
that operations data was collected, which is unlikely