* metrics: Add nil check for metricsHandler in AdminMetrics.serveHTTP
Prevents panic when the admin metrics endpoint is accessed before
the module is fully provisioned. Returns a proper API error instead
of crashing.
* admin: provision router modules before registering routes
Instead of adding a nil check for metricsHandler, address the root
cause by provisioning admin router modules before calling Routes().
This ensures all handler state is initialized before routes are
registered on the mux.
Merge newAdminHandler and provisionAdminRouters into a single step,
removing the two-phase setup where routes were registered first and
modules provisioned later. The AdminConfig.routers field is no longer
needed since provisioning happens inline.
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix: go fmt admin.go
---------
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* chore: ugh, lint fix...
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* more lint fixes
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* fix(provisioning): remove app from apps map when its provision failed
* Clean up failed app provisioning with defer
* fix(provisioning): record apps that failed to provision with their error
* save the error when an app fails to initialize and return this error when this app is requested by a module
---------
Co-authored-by: Matt Holt <mholt@users.noreply.github.com>
Co-authored-by: WeidiDeng <weidi_deng@icloud.com>
* events: Refactor; move Event into core, so core can emit events
Requires some slight trickery to invert dependencies. We can't have the caddy package import the caddyevents package, because caddyevents imports caddy. Interface to the rescue!
Also add two new events, experimentally: started, and stopping. At the request of a sponsor.
Also rename "Filesystems" to "FileSystems" to match Go convention (unrelated to events, was just bugging me when I noticed it).
* Coupla bug fixes
* lol whoops
* Add per host config
* Pass host label when option is enabled
* Test per host enabled
* metrics: scope metrics per loaded config
* doc and linter
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* inject the custom registry into the admin handler
Co-Authored-By: Dave Henderson <dhenderson@gmail.com>
* remove `TODO` comment
* fixes
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* refactor to delay metrics admin handler provision
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
Co-authored-by: Hussam Almarzooq <me@hussam.io>
Co-authored-by: Dave Henderson <dhenderson@gmail.com>
* Split `run` into a public `BuildContext` and a private part
`BuildContext` can be used to set up a caddy context from a config, but not start any listeners
or active components: The returned context has the configured apps provisioned, but otherwise is
inert.
This is EXPERIMENTAL: Minimally it's missing documentation and the example for how this can be
used to run unit tests.
* Use the config from the context
The config passed into `BuildContext` can be nil, in which case `BuildContext` will just make one
up that works. In either case that will end up in the finished context.
* Rename `BuildContext` to `ProvisionContext` to better match the function
* Hide the `replaceAdminServer` parts
The admin server is a global thing, and in the envisioned use case for `ProvisionContext`
shouldn't actually exist. Hide this detail in a private `provisionContext` instead, and
only expose it publicly with `replaceAdminServer` set to `false`.
This should reduce foot-shooting potential further; in addition the documentation comment
now clearly spells out that the exact interface and implementation details of `ProvisionContext`
are experimental and subject to change.
* use gofmput to format code
* use gci to format imports
* reconfigure gci
* linter autofixes
* rearrange imports a little
* export GOOS=windows golangci-lint run ./... --fix
* version: don't panic if read build info doesn't work
If `debug.ReadBuildInfo()` doesn't return the build information we
should not try to access it. Especially if users only want to build with
the `CustomVersion` we should not assume access to
`debug.ReadBuildInfo()`.
The build environment where this isn't available for me is when building
with bazel.
* exit early
* Allow version to be set manually
When Caddy is built from a release tarball (as downloaded from GitHub),
`caddy version` returns an empty string. This causes confusion for
downstream packagers.
With this commit, VersionString can be set with eg.
go build (...) -ldflags '-X (...).VersionString=v1.2.3'
Then the short form version will be "v1.2.3", and the full version
string will begin with "v1.2.3 ".
* Prefer embedded version, then CustomVersion
* Prefer "unknown" for full version over empty
Co-authored-by: Matthew Holt <mholt@users.noreply.github.com>
* caddyhttp: Honor grace period in background
This avoids blocking during config reloads.
* Don't quit process until servers shut down
* Make tests more likely to pass on fast CI (#5045)
* caddyhttp: Even faster shutdowns
Simultaneously shut down all HTTP servers, rather than one at a time.
In practice there usually won't be more than 1 that lingers. But this
code ensures that they all Shutdown() in their own goroutine
and then we wait for them at the end (if exiting).
We also wait for them to start up so we can be fairly confident the
shutdowns have begun; i.e. old servers no longer
accepting new connections.
* Fix comment typo
* Pull functions out of loop, for readability
Includes several breaking changes; code base updated accordingly.
- Added lots of context arguments
- Use fs.ErrNotExist
- Rename ACMEManager -> ACMEIssuer; CertificateManager -> Manager
* admin: Implement /pki/certificates/<id> API
* pki: Lower "skip_install_trust" log level to INFO
See https://github.com/caddyserver/caddy/issues/4058#issuecomment-976132935
It's not necessary to warn about this, because this was an option explicitly configured by the user. Still useful to log, but we don't need to be so loud about it.
* cmd: Export functions needed for PKI app, return API response to caller
* pki: Rewrite `caddy trust` command to use new admin endpoint instead
* pki: Rewrite `caddy untrust` command to support using admin endpoint
* Refactor cmd and pki packages for determining admin API endpoint
Br1an
Francis Lavoie
Salent Olivick
Alexandre Daubois
wangjingcun
Mohammed Al Sahaf
WeidiDeng
Laurin
Jimmy Lipham
Matt Holt
Andreas Kohn
Anton Kovalenko
a
Aziz Rmadi
Jacob Gadikian
Lukas Vogel
Adam Weinberger
jhwz
Kévin Dunglas
Noorain Panjwani
Денис Телюх
KallyDev