Note: This is a fork of the original goss-org/goss project created and maintained by @aelsabbahy. All original work remains under the Apache 2.0 license and full credit goes to the original author for building such a solid foundation. This fork exists to enable newer features and fixes to be developed and released. We are deeply grateful for the effort and care that went into the original project and aim to continue it in the same spirit.
Syver has since diverged from upstream. The work here is driven by client requirements rather than by tracking
goss-org/goss, so the two projects will continue to differ and Syver does not aim for release-for-release parity with upstream. Compatibility with the gossfile format is a separate matter and is maintained deliberately: existing gossfiles, environment variables and wrapper scripts keep working. See goss vs Syver for exactly what differs.
Note: For testing containers see the dsyver wrapper.
There are also wrapper scripts for Kubernetes (ksyver)
and Docker Compose (dcsyver). The goss-named
dgoss, kgoss and dcgoss still ship alongside them and work unchanged.
Note: For some Docker/Kubernetes healthcheck, health endpoint, and container ordering examples, see the Docker/Kubernetes simplified health checks blog post.
Syver is a YAML based serverspec alternative tool for validating a server's configuration. It eases the process of writing tests by allowing the user to generate tests from the current system state. Once the test suite is written they can be executed, waited-on, or served as a health endpoint.
- Syver is EASY!
- Syver is FAST! - small-medium test suites are near instantaneous
- Syver is SMALL! - <20MB single self-contained binary
Syver is the renamed continuation of the krameff/goss fork. Your gossfiles do not
need to change. gossfile:, goss.yaml, GOSS_* env vars, the dgoss/dcgoss/kgoss
wrappers and the -g flag all keep working.
See goss vs Syver for the full side-by-side comparison, including the single intentional breaking change.
Note: For macOS and Windows, see platform support.
Build from source or install release binaries — see installation.
This installs syver, the dsyver container wrapper,
and dgoss as a compatibility shim.
Download pre-built binaries and install wrappers as described in installation.
make buildAlternatively, you can build it with goreleaser. To
build a binary, use goreleaser build, and to only build for the same OS and
architecture as the machine you're building on, include the --single-target
flag. The --clean flag will clean up any existing builds, and --snapshot
will allow you to build against something other than a tag.
Here's an example:
$ goreleaser build --clean --single-target --snapshot
• skipping validate...
• cleaning distribution directory
• loading environment variables
• getting and validating git state
• ignoring errors because this is a snapshot error=git doesn't contain any tags - either add a tag or use --snapshot
• using tags previous= current=v0.0.0
• pipe skipped or partially skipped reason=disabled during snapshot mode
• parsing tag
• setting defaults
• partial
• snapshotting
• building snapshot... version=0.0.1-next
• running before hooks
• running hook=go mod tidy
• ensuring distribution directory
• setting up metadata
• writing release metadata
• loading go mod information
• build prerequisites
• building binaries
• partial build match=target=linux_arm64_v8.0
• building paths=cmd/syver binaries=syver target=linux_arm64_v8.0
• took: 31s
• writing artifacts metadata
• build succeeded after 31s
• thanks for using GoReleaser!
$ tree dist
dist
├── artifacts.json
├── binaries_linux_arm64_v8.0
│ └── syver <- your binary
├── config.yaml
└── metadata.json
2 directories, 4 filesPublished at syver.readthedocs.io, built
from docs/ in this repository. The site is built from main,
so it reflects the latest release rather than unreleased work on devel.
Release history records every release and the commit it points at. CHANGELOG.md has the detail of what changed in each.
Using the Syver container image
An initial set of tests can be derived from the system state by using the add or autoadd commands.
Let's write a simple sshd test using autoadd.
# Running it as root will allow it to also detect ports
$ sudo syver autoadd sshdOn spec filenames: syver looks for syver.yaml, syver.yml, goss.yaml
and goss.yml, in that order, and writes to whichever it found. When none
exist it creates syver.yaml. The goss-named files keep working indefinitely,
so an existing goss.yaml needs no migration; syver.yaml is simply the
preferred name for new work. Note this is the filename only. Inside the file
the import key stays gossfile: (syverfile: is accepted as an input alias).
Generated syver.yaml:
port:
tcp:22:
listening: true
ip:
- 0.0.0.0
pid:
- 1234
tcp6:22:
listening: true
ip:
- '::'
pid:
- 1234
service:
sshd:
enabled: true
running: true
user:
sshd:
exists: true
uid: 74
gid: 74
groups:
- sshd
home: /var/empty/sshd
shell: /sbin/nologin
group:
sshd:
exists: true
gid: 74
process:
sshd:
running: true
status:
- sleep
user:
- rootNow that we have a test suite, we can:
- Run it once
$ syver validate
...............
Total Duration: 0.021s # <- yeah, it's that fast..
Count: 15, Failed: 0- Edit it to use templates, and run with a vars file
syver --vars vars.yaml validate- keep running it until the system enters a valid state or we timeout
syver validate --retry-timeout 30s --sleep 1s- serve the tests as a health endpoint
$ syver serve &
$ curl localhost:8080/healthz
# JSON endpoint
$ syver serve --format json &
$ curl localhost:8080/healthz
# rspecish response via content negotiation
$ syver serve --format json &
$ curl -H "Accept: application/vnd.goss-rspecish" localhost:8080/healthzSyver files can be manually edited to improve readability and expressiveness of tests.
A Json draft 7 schema in
docs/schema.yaml
makes it easier to edit simple syver.yaml /
goss.yaml files in IDEs,
providing usual coding assistance such as inline documentation, completion and static analysis.
See #793 for screenshots.
For example, to configure the Json schema in JetBrains intellij IDEA, follow documented instructions, with arguments such as:
schema url=docs/schema.yaml(path from the repository root)schema version=Json schema version 7file path pattern=*/syver.yaml(add a second pattern for*/goss.yamlif you still have goss-named specs)
In addition, Syver files can also be further manually edited (without yet full json support) to use:
- Matchers and patterns
- Advanced Matchers
- Templates
titleandmeta(arbitrary data) attributes are persisted when adding other resources withsyver add
A typo'd top-level key (writing prot: instead of port:, for example) no
longer validates silently: syver logs a [WARN] for it, with a suggestion
when it can make one. See Unknown top-level keys.
Some examples:
user:
sshd:
title: UID must be between 50-100, GID doesn't matter. home is flexible
meta:
desc: Ensure sshd is enabled and running since it's needed for system management
sev: 5
exists: true
uid:
# Validate that UID is between 50 and 100
and:
gt: 50
lt: 100
home:
# Home can be any of the following
or:
- /var/empty/sshd
- /var/run/sshd
package:
kernel:
installed: true
versions:
# Must have 3 kernels and none of them can be 4.4.0
and:
- have-len: 3
- not:
contain-element: 4.4.0
# Loaded from --vars YAML/JSON file
{{.Vars.package}}:
installed: true
{{if eq .Env.OS "centos"}}
# This test is only when $OS environment variable is set to "centos"
libselinux:
installed: true
{{end}}
Spec files with templates can still be validated through the Json schema after being rendered
using the syver render command. See example below
$ cd docs
$ syver --vars ./vars.yaml render > rendered_syver.yaml
# proceed with json schema validation of rendered_syver.yaml in your favorite IDE
# or in one of the Json schema validator listed in https://json-schema.org/implementations.html
# The following example is for a Linux AMD64 host
$ curl -LO https://github.com/neilpa/yajsv/releases/download/v1.4.1/yajsv.linux.amd64
$ chmod a+x yajsv.linux.amd64
$ sudo mv yajsv.linux.amd64 /usr/sbin/yajsv
$ yajsv -s schema.yaml rendered_syver.yaml
rendered_syver.yaml: passFull list of available Json schema validators can be found in https://json-schema.org/implementations.html#validator-command%20line
Run lightweight discovery checks before the main suite and use the results in templates, or declare
depends-on to skip dependents when a prerequisite fails. See
Discovery and
Test dependencies.
# Pre-run discovery, then validate main gossfile (preferred)
syver validate -g goss.yml --discover discovery.yaml
# Or inline discovery: in the same gossfile
syver validate -g goss-inline.yml
# Export discovery results for external tooling (unchanged)
syver validate -g discovery.yaml --format discoveryThese examples pass goss-named files because that is what the checked-in
fixtures are called. goss.yaml and goss.yml are still accepted, and
always will be. -g takes any path, and a spec found by name is resolved
across all four of syver.yaml, syver.yml, goss.yaml, goss.yml.
Fixtures: integration-tests/syver/examples/discovery/
- package - add new package
- file - add new file
- addr - add new remote address:port - ex: google.com:80
- port - add new listening [protocol]:port - ex: 80 or udp:123
- service - add new service
- user - add new user
- group - add new group
- command - add new command
- dns - add new dns
- process - add new process name
- registry - add new Windows registry key or value (Windows only)
- kernel-param - add new kernel-param
- mount - add new mount
- interface - add new network interface
- http - add new network http url with proxy support
- syver (alias: goss) - add new syver file, it will be imported from this one
- matching - test for matches in supplied content
- rspecish - (default) Similar to rspec output
- documentation - Verbose test results
- json - JSON, detailed test result
- structured - JSON like
json, plus asummaryobject and a human-readablesummary-lineon every result - tap - TAP style
- junit - JUnit style
- nagios - Nagios/Sensu compatible output /w exit code 2 for failures.
- prometheus - Prometheus compatible output.
- silent - No output. Avoids exposing system information (e.g. when serving tests as a healthcheck endpoint).
These are third-party integrations written for upstream goss. They are listed
because they still work, but each one invokes a binary named goss, and
install.sh installs syver, dsyver and dgoss, but no goss. To use any of
them, put a goss on your PATH pointing at syver:
sudo ln -s "$(command -v syver)" /usr/local/bin/gossThe gossfiles these tools generate and consume need no changes; only the binary name differs. None of them are maintained by this project.
- goss-ansible - Ansible module for Goss.
- degoss - Ansible role for installing, running, and removing Goss in a single go.
- ansible-goss-install - Ansible role for installing Goss (option for install as user or root)
- kitchen-goss - A test-kitchen verifier plugin for Goss.
- goss-fpm-files - Might be useful for building goss system packages.
- packer-provisioner-goss - A packer plugin to run Goss as a provision step.
- gossboss - Collect and view aggregated Goss test results from multiple remote Goss servers.
syver works well on Linux, but support on Windows & macOS is alpha. See platform support.
The following tests have limitations.
Package:
- rpm
- deb
- Alpine apk
- pacman
Service:
- systemd
- sysV init
- OpenRC init
- Upstart
Port:
- Port state is read from
/proc/net/{tcp,udp,tcp6,udp6}on Linux, where it is fully supported. It is not implemented on macOS or Windows: measured on Windows Server 2025, every assertion returns "not implemented yet". See platform support, which is the authoritative per-resource matrix. - On Linux, if one of those files exists but contains a line syver can't parse
(an unexpected IP/port/uid encoding, typically from a non-standard procfs,
e.g. inside certain containers or network namespaces), the affected
portresource now fails with an explicitError:block in the output instead of silently reporting the port as not listening. To investigate: note which protocol failed from the resource id (tcp,tcp6,udp,udp6), then inspect the corresponding file directly, e.g.cat /proc/net/tcp6, and compare its columns against a working host. A missing or unreadable file is not an error case -- it's treated as "no ports" for that protocol.