[Feature Request] Add support for customizable Type value in heka encoders

[Matt Simonin]

In the current implementation, Kolla uses heka to sends parsed log lines to elasticsearch output.
If one sends differents Openstack instances logs into the same elasticsearch instance, I don't see any obvious way to differentiate the source of the logs.

I've slightly modified all the decoders to allow the Type to be passed through the configuration like this :

[openstack_log_decoder]
type = "SandboxDecoder"
filename = "lua_decoders/os_openstack_log.lua"

[openstack_log_decoder.config]
type = "my custom type"

Each decoder have to be slighly modified to read the value from the config. The value can be defaulted to the current hard-coded value to ensure backward compatibility [1].

Let me know, if you think it's worth a review on gerrit.

Otherwise if there's another way to differentiate log provenance in a single elasticsearch instance, I'll be happy to learn it.

[1]: https://github.com/openstack/kolla/blob/789f50b2f3bc733015d11287054b63db7c01c03a/docker/heka/plugins/decoders/os_openstack_log.lua#L24

[Duong Ha-Quang (duonghq)]

Due to deprecation of Heka, Kolla is moving to another logging engine. So any improvement is still welcome but not we should not put much effort on this.

Anyway, if you are not sure about a review on gerrit, it's always worth it (don't forget tag it with appropriate bug id)

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/381764

[OpenStack Infra (hudson-openstack)]

Change abandoned by Matthieu Simonin (<email address hidden>) on branch: master
Review: https://review.openstack.org/381764
Reason: Heka will be deprecated in ocata, i'm withdrawing this patch.

[Christian Berendt (berendt)]

Heka will be deprecated in Ocata.