syslog-ng: the listener it never configures, and the throughput numbers it never qualifies
syslog-ng is an enhanced log daemon, supporting a wide range of input and output methods: syslog, unstructured text, queueing, SQL & NoSQL.
At a glance
- What is it?
- An enhanced log daemon written in C, quickstart examples that depend on a library file you must install separately, and a performance section whose two figures differ by more than an order of magnitude. A look at what the configuration does not tell you.
- Who is it for?
- Judgment: syslog-ng earns its place on a host that has to reshape logs before they land, and the parsers, templates and destinations it offers go well past what a plain relay does. Two reservations before you commit.
- Can I use it commercially?
- Check first. The repository uses a licence we do not classify automatically, so read its LICENSE file before any commercial use.
- Is it still maintained?
- Yes. The repository last received commits 4 days ago.
- What is it written in?
- Mainly C, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.
Star history
- syslog-ng/syslog-ng2,387 ★ · +10
| Repository | 2026-09-30 | 2026-11-15 |
|---|---|---|
| syslog-ng/syslog-ng | 2.4k | 2.4k |
10 Stars in the last 46 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
Every quickstart config opens by including scl.conf, so none of them is self contained
All three quickstart configurations begin the same way, with a version selector that names no version and an include of a library file:
@version: current
@include "scl.conf"
log {
source { system(); };
destination { file("/var/log/syslog"); };
};Two things follow from those two lines. The `current` in `@version` is a symbolic keyword rather than a number, so it resolves against whatever library definitions the installation provides instead of pinning the behaviour you are copying. And `scl.conf` is the compatibility library that lives in the `scl/` directory of the source tree, which means a configuration lifted out of these pages does not run until that file is on the include path of the installation you lifted it into. None of the three examples says where to put it.
A smaller detail worth knowing before you copy anything: each of these fences is tagged with a space before the language name, so the blocks are not highlighted as configuration wherever they render.
The third config is the only one that flattens the JSON it advertises
The page demonstrates a round trip. The submission is JSON behind a namespace prefix:
$ logger '@cim: {"name1":"value1", "name2":"value2"}'and the message that comes out the other side is meant to read as key=value pairs:
name1=value1 name2=value2Bridging those two takes the template on the third destination, and that template is the only substantive difference between the third configuration and the first:
@version: current
@include "scl.conf"
log {
source { system(); };
destination { file("/var/log/app.log" template("$(format-welf --subkeys .cim.)\n")); };
};The first two examples set no template at all. Copy either of those and the message reaches the destination with its `@cim:` prefix and its braces intact, which is the right behaviour if you wanted the raw input and the wrong one if you were counting on the flattened output the third example advertises.
Adding network() switches on a listener the config file never names
The second example adds a single line to the source and with it a network listener the configuration never spells out:
@version: current
@include "scl.conf"
log {
source {
system();
network();
};
destination { file("/var/log/syslog"); };
};The sentence above that block says the configuration processes logs from the network, with TCP port 514 by default. No port, no protocol and no bind address appear anywhere inside it. The address the daemon listens on therefore lives in the defaults compiled into the binary rather than in the file you edit, which makes the first task on any unexpected socket behaviour finding out what those defaults currently resolve to on your build.
The local source has the same shape. `system()` takes everything arriving on the system socket, whether an application wrote it there or systemd forwarded it, and where that boundary sits is stated in prose rather than expressed in the configuration.
The performance section quotes 600-800k and several tens of thousands with no conditions
The performance section is short and its two numbers do not sit comfortably together. The claim is that syslog-ng provides performance levels comparable to a large cluster while running on a single node, and that in the simplest use case it scales up to 600-800k messages per second. The following sentence then says classification, parsing and filtering still produce several tens of thousands of messages per second.
That is a drop of more than an order of magnitude between the simplest case and the case a real deployment lives in. Nothing in the section says what hardware either figure came from, how large the messages were, how many were in flight at once, or whether drops were counted. The cluster being compared against is not named either, so there is nothing to compare against.
The figures are worth having and not worth planning on. Size a host for the lower number, since that is the one describing work the tool exists to do, and treat the higher one as the cost of the pass-through case.
Build dependencies sit below what the comments say is needed
The Python build requirements carry their own commentary, and the commentary is where the constraint lives. Three entries are deliberately held below the version the project says it wants:
boto3>=1.37.38 # we would need >= 1.40.44 (to sync with PE), but highest available on kira - ubuntu focal
urllib3>=1.26 # we would need >= 2.6 (because high vulnerability), but botocore requires <1.27,>=1.25.4
tornado>= 6.4.2 # we would need >= 6.5 (because high vulnerability), but highest available on kira - ubuntu focalThe stated reason is the build host, an Ubuntu Focal machine on which the newest wheels are not published. For two of the three the reason given is a security one. urllib3 is wanted at 2.6 or above because of a high vulnerability and cannot go there because botocore constrains it below 1.27. tornado is wanted at 6.5 or above for the same kind of reason and stays at 6.4.2 for want of a build that host can resolve.
The phrase repeated across the file is a sync with PE, which points at the commercial edition named in this same repository. Open source build reproducibility is defined here against a product that is not open source, with two of the pins held back by a machine that can only see as far as Focal.
CMake, autotools and a second pkg-config file share one root
The repository root carries the artefacts of three build conventions at once. `CMakeLists.txt` sits beside `configure.ac`, `Makefile.am`, `autogen.sh`, `sub-configure.sh`, an `m4/` directory and an `Mk/` directory, so this is a CMake project that also answers to autotools and to BSD style make includes.
Next to those sit two pkg-config templates, `syslog-ng.pc.in` and `syslog-ng-native-connector.pc.in`, each with a `.cmake` companion. A native connector is therefore installed as its own findable package, and the source build section of the README never mentions it, which leaves a consumer who finds the interface advertised with nothing explaining what it is for.
What that section does warn about is that some functionality is compiled only when the required development libraries are present, and that the configure script prints a summary of enabled features when it finishes. That summary is the only record of what a particular build is capable of, which makes it worth keeping rather than scrolling past.
The contact paths still carry the domain of the company acquired in 2018
The sponsorship section tells the project's history in two sentences. Balabit was the original commercial sponsor and was acquired by One Identity in 2018, which offers a commercial edition called the syslog-ng Premium Edition. Axoflow is the company of Balazs Scheidler, the original creator and main developer.
The contact paths carry the older brand. The mailing list is hosted at lists.balabit.hu, a domain belonging to the company that was acquired, while the feedback address sits at syslog-ng.org, the project site is syslog-ng.com, and the package repository is at ose-repo.syslog-ng.com. Four addresses for one project, one of them tied to an owner that no longer exists, and none of them the domain in the repository homepage field.
The deployment model is stated in the same breath: development is described as volunteer work, reached through the GitHub project page, a gitter channel and that mailing list.
The licence field is unresolved, the default branch is develop, and the badges are blank
Three pieces of repository metadata are worth settling before relying on them. The first is the licence. The licence field is left unresolved, while the root carries `COPYING`, `GPL.txt` and `LGPL.txt` side by side. Nothing in the repository chooses between them for you, so the terms that apply depend on which file a reader opens.
The second is the default branch, which is `develop` rather than `main` or `master`. That is the branch a clone checks out, and it sits ahead of the newest tag: the latest release is syslog-ng-4.12.0 from 16 June 2026, and the branch itself was pushed on 2 October 2026.
The third is the top of the README, which is four workflow badges with no alt text and no visible label: a development shell workflow, a nightly release workflow, a packaging workflow and a container image workflow. Useful signals, none of them labelled, so nothing at a glance tells you which of them gates a change.
Editorial conclusion
Judgment: syslog-ng earns its place on a host that has to reshape logs before they land, and the parsers, templates and destinations it offers go well past what a plain relay does. Two reservations before you commit. The published throughput figures are not a sizing input, so plan capacity for the tens of thousands figure rather than the 600-800k one and measure on your own messages. And the licence is unresolved in the repository metadata while GPL and LGPL texts both sit at the root, so settle which terms apply before this reaches a build you have to defend. The configuration examples are worth reading closely for a third reason: the port the daemon listens on is not in the config file.
Frequently asked questions
What is syslog-ng used for?
It is an enhanced log daemon that receives and sends RFC3164 and RFC5424 style syslog messages as well as JSON formatted messages and unstructured data. It classifies and structures logs with built-in parsers such as csv-parser(), db-parser() and kv-parser(), then hands them on to files, message queues like AMQP, databases like PostgreSQL and MongoDB, or tools like Elasticsearch, Kafka and Hadoop.
Is syslog-ng free?
The licence situation needs reading rather than assuming: the licence field on this repository is left unresolved, while COPYING, GPL.txt and LGPL.txt all sit at the root. One Identity, which acquired the original sponsor Balabit in 2018, offers a commercial edition called the syslog-ng Premium Edition. Packages from the project's own APT repository are provided as is, without warranty of any kind, on a best-effort level.
Which is better, rsyslog or syslog-ng?
This repository makes no such comparison, so it cannot answer that. What it does state is its own scope, which covers classification, parsing and filtering of logs rather than pass-through relay alone, and its own performance claims of 600-800k messages per second in the simplest use case and several tens of thousands once classification and parsing are involved, without naming hardware or conditions for either figure.
how to install syslog ng on ubuntu 24.04
From binaries, run `apt install syslog-ng` as root against the project's APT repository, which publishes a stable and a nightly channel per distribution, with separate sources.list component names for x86-64 and arm64. The visible rows of the support table pair Ubuntu 26.04 with ubuntu-resolute, 25.04 with ubuntu-plucky and 24.04 with ubuntu-noble. From source, run ./autogen.sh first if you cloned from git, then ./configure && make && make install.
what is syslog ng ose
The open source edition is this repository, and its own package host is ose-repo.syslog-ng.com, which serves the Debian and Ubuntu stable and nightly channels. The commercial counterpart named in the same file is the syslog-ng Premium Edition offered by One Identity, and Axoflow is the company of Balazs Scheidler, the original creator and main developer.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/syslog-ng-syslog-ng)Star history chart for your README
[](https://hysenlabs.com/projects/syslog-ng-syslog-ng)Was this useful?