Good recommendation on how to stream log data into Amazon Elastic Search?

1,661 views
Skip to first unread message

Marco Pas

unread,
Jun 2, 2017, 3:32:28 AM6/2/17
to Fluentd Google Group
Does anyone have a good stable recommendation to stream log data into Amazon Elastic Search? We are using the aws-elasticsearch-service plugin but we keep running into issues. Like required automatic restarts for the aggregator. (https://github.com/atomita/fluent-plugin-aws-elasticsearch-service)

David Wood

unread,
Jun 3, 2017, 8:12:23 AM6/3/17
to Fluentd Google Group

Marco Pas

unread,
Jun 4, 2017, 5:46:57 AM6/4/17
to flu...@googlegroups.com, David Wood
Hi David,

do you have an example of a configuration for plugin configuration for AWS Elasticsearch that seems to work? We now need to reboot the aggregators very day to keep them going and this is causing problems.
--
You received this message because you are subscribed to a topic in the Google Groups "Fluentd Google Group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/fluentd/uW87VAOqxeE/unsubscribe.
To unsubscribe from this group and all its topics, send an email to fluentd+u...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

David Wood

unread,
Jun 4, 2017, 7:41:04 AM6/4/17
to Fluentd Google Group, ya...@yankeefrog.com
Hi,


I instead use https://github.com/uken/fluent-plugin-elasticsearch.  It doesn't support Amazon signing so you'll need an IP access policy.  You'd need to ensure you have these settings:

resurrect_after 5s
reload_connections false
buffer_chunk_limit 8m

The buffer_chunk_limit could be larger depending on the instance size you select.  The link I provided earlier has more information.

David

David Wood

unread,
Jun 4, 2017, 7:49:29 AM6/4/17
to Fluentd Google Group, ya...@yankeefrog.com

Marco Pas

unread,
Jun 6, 2017, 4:53:50 AM6/6/17
to flu...@googlegroups.com, David Wood
Hi David,

when using these settings do you still need te restart the Fluented container now and then? We ran into the issues that a restart of Fluent is needed on a scheduled basis. When we did not do this the inflow would stop into ES.

If an IP access policy is used is there still a requirement to do the signing?

- Marco

David Wood

unread,
Jun 6, 2017, 8:23:57 AM6/6/17
to Fluentd Google Group, ya...@yankeefrog.com
when using these settings do you still need te restart the Fluented container now and then?
No - it runs for months without any issue. 
 
If an IP access policy is used is there still a requirement to do the signing?
No - no need for both.

David

Marco Pas

unread,
Jun 15, 2017, 10:41:58 AM6/15/17
to Fluentd Google Group, ya...@yankeefrog.com
I am for sure going to try this out. If i point this plugin to AWS ES what are the settings for the host / port?

David Wood

unread,
Jun 15, 2017, 10:45:58 AM6/15/17
to Fluentd Google Group, ya...@yankeefrog.com
The port is 443.  The host is whatever is listed as the "endpoint" in the AWS console (search-XXXXXXXXX.es.amazonaws.com).

David

Marco Pas

unread,
Jun 15, 2017, 10:46:28 AM6/15/17
to Fluentd Google Group, ya...@yankeefrog.com
Would you be able to share you config when using this plugin and pointing it to AWS ES?

I am trying to implement the same and an example how your succesful config is would be great!

@type elasticsearch
logstash_format true
flush_interval 60s

host <points to??>
port <is the port needed ??>
index_name fluentd
type_name fluentd

resurrect_after 5s
reload_connections false
buffer_chunk_limit 8m


On Tuesday, June 6, 2017 at 2:23:57 PM UTC+2, David Wood wrote:

David Wood

unread,
Jun 15, 2017, 10:56:44 AM6/15/17
to Fluentd Google Group, ya...@yankeefrog.com
Sure:
      @type elasticsearch
      host XXXXX
      port 443
      scheme https
      request_timeout 30s
      resurrect_after 5
      # -new-connection-from-pool-error/36252/6
      reload_connections false
      logstash_format true
      logstash_prefix syslog
      # @timestamp: use event time, not time of indexing
      time_key time
      include_tag_key true
      # aes-limits.html
      buffer_chunk_limit 8m
      buffer_queue_limit 2048    # 8m * 2048 = 16GB

David Wood

unread,
Jun 21, 2017, 2:20:19 PM6/21/17
to Fluentd Google Group, ya...@yankeefrog.com
 How'd you make out?  David



Marco Pas

unread,
Jul 3, 2017, 12:41:28 PM7/3/17
to flu...@googlegroups.com
Hi David, 

apologies for the delay. We have just integrated it into our infrastructure and until now it seems to work great. No more hick ups and data is nicely flowing into AWS ES. Thanks a lot for your help!!!!

On Wed, Jun 21, 2017 at 8:20 PM David Wood <ya...@yankeefrog.com> wrote:
 How'd you make out?  David



Reply all
Reply to author
Forward
0 new messages