Increase logging description - errors/warnings from processing don't describe source

[epcim]

Check CONTRIBUTING guideline first and here is the list to help us investigate the problem.

• fluentd or td-agent version.
  td-agent, 0.14.20

• Environment information, e.g. OS.
  container, k8s, https://github.com/Mirantis/k8s-apps/tree/master/charts/fluentd

• Your configuration
  N/A

I see following error in log, but no idea what filter/record/source/label it;s related:

2017-11-03 09:40:36 +0000 [warn]: #0 dump an error event: error_class=NoMethodError error="undefined method `merge!' for \"o\":String"
tag="openstack.neutron.server" time=#<Fluent::EventTime:0x00005587b65c65c8 @sec=1509702036, @nsec=192057421> record="o"

Another ocurence is

2017-11-03 09:40:36 +0000 [warn]: #0 dump an error event: error_class=TypeError error="can't dup Fixnum" tag="openstack.neutron.server"
 time=1 record=0
2017-11-03 09:40:36 +0000 [warn]: #0 dump an error event: error_class=TypeError error="can't dup Fixnum" tag="openstack.neutron.server"
 time=1 record=1
2017-11-03 09:40:36 +0000 [warn]: #0 dump an error event: error_class=TypeError error="can't dup Fixnum" tag="openstack.neutron.server"
 time=#<Fluent::EventTime:0x00005587b65c67f8 @sec=1509702036, @nsec=192047148> record=0
2017-11-03 09:40:36 +0000 [warn]: #0 dump an error event: error_class=TypeError error="can't dup Fixnum" tag="openstack.neutron.server"
 time=1 record=1

Obviously it's just wrong assumption about the value type and applying of a wrong method.

I have tried FLUENTD_OPT="-vv" but that didtn gave me much more (assuming then I am even more overloaded with too much details anyway). Each Error/Warning whould have had full 5W (where,what,why,when,who)...

If possible, please guide me what I can do more to resolve these two issues.

[epcim]

I have one more nice example

2017-11-03 14:40:27 +0000 [warn]: #0 dump an error event: error_class=NoMethodError error="undefined method `merge!' for [49, 49, 57]:A
rray" tag="openstack.neutron.server" time=#<Fluent::EventTime:0x0000564713af2b10 @sec=1509720027, @nsec=182093896> record=[49, 49, 57]

Please let me know how I may help you to fix that. Is that an obvious bug or do we have to change something in larger scope.

[repeatedly]

Hmm... I see. Showing stacktrace is better in emit, not only error message.
I will add it.