740d448983
So far we've been using Go's native time.Time for anything related to sample timestamps. Since the range of time.Time is much bigger than what we need, this has created two problems: - there could be time.Time values which were out of the range/precision of the time type that we persist to disk, therefore causing incorrectly ordered keys. One bug caused by this was: https://github.com/prometheus/prometheus/issues/367 It would be good to use a timestamp type that's more closely aligned with what the underlying storage supports. - sizeof(time.Time) is 192, while Prometheus should be ok with a single 64-bit Unix timestamp (possibly even a 32-bit one). Since we store samples in large numbers, this seriously affects memory usage. Furthermore, copying/working with the data will be faster if it's smaller. *MEMORY USAGE RESULTS* Initial memory usage comparisons for a running Prometheus with 1 timeseries and 100,000 samples show roughly a 13% decrease in total (VIRT) memory usage. In my tests, this advantage for some reason decreased a bit the more samples the timeseries had (to 5-7% for millions of samples). This I can't fully explain, but perhaps garbage collection issues were involved. *WHEN TO USE THE NEW TIMESTAMP TYPE* The new clientmodel.Timestamp type should be used whenever time calculations are either directly or indirectly related to sample timestamps. For example: - the timestamp of a sample itself - all kinds of watermarks - anything that may become or is compared to a sample timestamp (like the timestamp passed into Target.Scrape()). When to still use time.Time: - for measuring durations/times not related to sample timestamps, like duration telemetry exporting, timers that indicate how frequently to execute some action, etc. *NOTE ON OPERATOR OPTIMIZATION TESTS* We don't use operator optimization code anymore, but it still lives in the code as dead code. It still has tests, but I couldn't get all of them to pass with the new timestamp format. I commented out the failing cases for now, but we should probably remove the dead code soon. I just didn't want to do that in the same change as this. Change-Id: I821787414b0debe85c9fffaeb57abd453727af0f |
||
|---|---|---|
| .build | ||
| coding | ||
| config | ||
| documentation/examples | ||
| model | ||
| notification | ||
| retrieval | ||
| rules | ||
| stats | ||
| storage | ||
| tools | ||
| utility | ||
| web | ||
| .gitignore | ||
| .pkgignore | ||
| .travis.yml | ||
| build_info.go | ||
| CONTRIBUTING.md | ||
| CONTRIBUTORS.md | ||
| Dockerfile | ||
| LICENSE | ||
| main.go | ||
| Makefile | ||
| Makefile.INCLUDE | ||
| MANIFEST | ||
| README.md | ||
| tests-for-die-in-a-fire-travis.sh | ||
Prometheus
Bedecke deinen Himmel, Zeus! A new kid is in town.
Prometheus is a generic time series collection and computation server that is useful in the following fields:
- Industrial Experimentation / Real-Time Behavioral Validation / Software Release Qualification
- Econometric and Natural Sciences
- Operational Concerns and Monitoring
The system is designed to collect telemetry from named targets on given intervals, evaluate rule expressions, display the results, and trigger an action if some condition is observed to be true.
Prerequisites
If you read below in the Getting Started section, the build infrastructure will take care of the following things for you in most cases:
- Go 1.1.
- LevelDB: https://code.google.com/p/leveldb/.
- Protocol Buffers Compiler: http://code.google.com/p/protobuf/.
- goprotobuf: the code generator and runtime library: http://code.google.com/p/goprotobuf/.
- Levigo, a Go-wrapper around LevelDB's C library: https://github.com/jmhodges/levigo.
- GoRest, a RESTful style web-services framework: http://code.google.com/p/gorest/.
- Prometheus Client, Prometheus in Prometheus https://github.com/prometheus/client_golang.
- Snappy, a compression library for LevelDB and Levigo http://code.google.com/p/snappy/.
Getting Started
For basic help how to get started:
- The source code is periodically indexed: Prometheus Core.
- For UNIX-like environment users, please consult the Travis CI configuration in .travis.yml and Makefile.
- All of the core developers are accessible via the Prometheus Developers Mailinglist.
General
For first time users, simply run the following:
$ make
$ ARGUMENTS="-configFile=documentation/examples/prometheus.conf" make run
${ARGUMENTS} is passed verbatim into the makefile and thusly Prometheus as
$(ARGUMENTS). This is useful for quick one-off invocations and smoke
testing.
If you run into problems, try the following:
$ SILENCE_THIRD_PARTY_BUILDS=false make
Upon having a satisfactory build, it's possible to create an artifact for end-user distribution:
$ make package
$ find build/package
build/package will be sufficient for whatever archiving mechanism you
choose. The important thing to note is that Go presently does not
staticly link against C dependency libraries, so including the lib
directory is paramount. Providing LD_LIBRARY_PATH or
DYLD_LIBRARY_PATH in a scaffolding shell script is advised.
Problems
If at any point you run into an error with the make build system in terms of
its not properly scaffolding things on a given environment, please file a bug or
open a pull request with your changes if you can fix it yourself.
Please note that we're explicitly shooting for stable runtime environments and not the latest-whiz bang releases; thusly, we ask you to provide ample architecture and release identification remarks for us.
Testing
$ make test
Packaging
$ make package
Race Detector
Go 1.1 includes a race detector which can be enabled at build time. Here's how to use it with Prometheus (assumes that you've already run a successful build).
To run the tests with race detection:
$ GORACE="log_path=/tmp/foo" go test -race ./...
To run the server with race detection:
$ go build -race .
$ GORACE="log_path=/tmp/foo" ./prometheus
Contributing
Refer to CONTRIBUTING.md
License
Apache License 2.0