Open johnarnold opened 8 years ago
I also have to say losing this is a problem, as I don't see a solution other than this for my current issue: I have docker containers write out their logs via the json-file driver (anything else and it's a lot of junk in the syslog or just hard to seperate logs from multiple containers), but since filebeats collects them as json, it's too late to multiline at input, I have to do that as a filter. Unless I'm missing something important?
Hi,
The documentation states that this filter is being deprecated in favor of the multiline codec. However, multiline codec doesn't support custom stream_identity's and therefore can't do transaction id style streams over a single input (a great example is Cisco ACS AAA logs).
From the wayback machine, here's some examples of people trying to do this. https://logstash.jira.com/browse/LOGSTASH-1785 It also shows up in google and other open issues / JIRA's.
If you want to add a regex/grok to the multiline codec to support custom stream_identity that would probably suffice for most -- although it's not as "clean" as the filter implementation against fully formed events.
There's an interesting reference to "sub-stream patterns" here which is a similar idea: https://github.com/logstash-plugins/logstash-codec-multiline/issues/22
Is there a way to do transaction id style multiline in 5.0.0?
Thanks, John