With no differentiator the Logger value will be the name you give the
LogstreamerInput plugin, which has to be unique. The only way a single
LogstreamerInput can generate multiple streams is by specifying a
differentiator, in which case the Logger value will get the differentiator's
name for each stream.
The only possible issue I can see is that if you specify the same
differentiator name structure in two different LogstreamerInput configurations,
Heka won't notice and you'll get name collision. So don't do that, and you
should be fine.
-r
On 02/08/2016 01:04 PM, Eli Flesher wrote:
Yeah, exactly right, I just need a unique identifier to the stream.
Right now, I’m using the Logger in a way similar to this:
record := LogAgentRecord{
...
Path: pack.Message.Logger,
Events: payload,
Datetime: ts,
...
}
However, what get’s encoded when that Struct is marshaled to json is:
"path":”LogstreamerInput"
However, I don’t have a differentiator specified nor do I have more than
one LogStreamerInput configured while testing. Let me try both of these
and I’ll update the thread.
Any other suggestions would be appreciated as well.
Thanks,
Eli
On Mon, Feb 8, 2016 at 12:25 PM Rob Miller <[email protected]
<mailto:[email protected]>> wrote:
Hrm. File names are tricky, b/c the LogstreamerInput usually thinks in
terms of log *streams*, not files, so the name of the file that a
particular record came from isn't readily available.
I suspect that you don't actually need the file, though, but that some
unique identifier for your stream is actually enough. If that's the
case, you should check out the `Logger` value on your generated
messages. That will contain your differentiator value, if you have a
differentiator specified. If not, it should at least be unique for each
LogstreamerInput in use.
If that's not good enough, you might have to do something yourself in
the decoding layer. Hopefully Logger will suffice, though.
-r
On 02/08/2016 11:33 AM, Eli Flesher wrote:
> Hello Everyone,
>
> We’re exploring using Heka internally as a daemon to push up
> periodically written files and submit changes to an API for
consumption.
>
> As part of this project, I’ve written an Encoder that formats
messages
> from several LogStreamerInputs into our message format before
going to
> an Output plugin that submits the messages.
>
> I’ve run into one problem. I need to know the file name being
monitored
> in order to determine parts of the message body. Looking at the
> LogStreamerInput code, I can’t see a way that this is exposed,
but I’m
> also very new to Go.
>
> Can anyone guide me on how I might accomplish this?
>
>
> Eli
> --
> —
> *Elijah Flesher* | *Lyft* <http://lyft.me/> | /Software Engineer/
> 206.661.4697 | @eliflesher
>
>
> _______________________________________________
> Heka mailing list
> [email protected] <mailto:[email protected]>
> https://mail.mozilla.org/listinfo/heka
>
--
—
*Elijah Flesher* | *Lyft* <http://lyft.me/> | /Software Engineer/
206.661.4697 | @eliflesher
_______________________________________________
Heka mailing list
[email protected]
https://mail.mozilla.org/listinfo/heka
_______________________________________________
Heka mailing list
[email protected]
https://mail.mozilla.org/listinfo/heka
