mtail - extract internal monitoring data from application logs for collection into a timeseries database
mtail
is a tool for extracting metrics from application logs to be exported
into a timeseries database or timeseries calculator for alerting and
dashboarding.
It fills a monitoring niche by being the glue between applications that do not export their own internal state (other than via logs) and existing monitoring systems, such that system operators do not need to patch those applications to instrument them or writing custom extraction code for every such application.
The extraction is controlled by mtail programs which define patterns and actions:
# simple line counter
counter lines_total
/$/ {
lines_total++
}
Metrics are exported for scraping by a collector as JSON or Prometheus format over HTTP, or can be periodically sent to a collectd, StatsD, or Graphite collector socket.
Read the programming guide if you want to learn how to write mtail programs.
Ask general questions on the users mailing list: https://groups.google.com/g/mtail-users
There are various ways of installing mtail.
Precompiled binaries for released versions are available in the Releases page on Github. Using the latest production release binary is the recommended way of installing mtail.
Windows, OSX and Linux binaries are available.
The simplest way to get mtail
is to go get
it directly.
go get github.com/google/mtail/cmd/mtail
This assumes you have a working Go environment with a recent Go version. Usually mtail is tested to work with the last two minor versions (e.g. Go 1.12 and Go 1.11).
If you want to fetch everything, you need to turn on Go Modules to succeed because of the way Go Modules have changed the way go get treats source trees with no Go code at the top level.
GO111MODULE=on go get -u github.com/google/mtail
cd $GOPATH/src/github.com/google/mtail
make install
If you develop the compiler you will need some additional tools
like goyacc
to be able to rebuild the parser.
See the Build instructions for more details.
A Dockerfile
is included in this repository for local development as an
alternative to installing Go in your environment, and takes care of all the
build dependency installation, if you don't care for that.
mtail
works best when it paired with a timeseries-based calculator and
alerting tool, like Prometheus.
So what you do is you take the metrics from the log files and you bring them down to the monitoring system?
It deals with the instrumentation so the engineers don't have to! It has the extraction skills! It is good at dealing with log files!!
Full documentation at http://google.github.io/mtail/
Read more about writing mtail
programs:
Read more about hacking on mtail
Read more about deploying mtail
and your programs in a monitoring environment
- Deploying
- Interoperability with other systems
- Troubleshooting
- FAQ
After that, if you have any questions, please email (and optionally join) the mailing list: https://groups.google.com/forum/#!forum/mtail-users or file a new issue.