* reverse proxy: rewrite requests and responses for websocket over http2
* delete protocol pseudo-header
* modify cloned requests
* set request variable to track if it's a h2 websocket
* use request bodu
* rewrite request body
* use WebSocket instead of Websocket in the headers
* use logger check for zap loggers
* fix lint
* fix: handle "request body too large" error using type assertion
* fix: address overlooked nil check for MaxBytesError
* fix: replace type assertion with errors.As() for MaxBytesError
i.e. Revert commit f5dce84a70
Two years ago, the patch in #4952 was a seemingly necessary way to fix an issue (sort of an edge case), but it broke other more common use cases (see #6666).
Now, as of #6669, it seems like the original issue can no longer be replicated, so we are reverting that patch, because it was incorrect anyway.
If it turns out the original issue returns, a more proper patch may be in #6669 (even if used as a baseline for a future fix). A potential future fix could be an opt-in setting.
* Allow 0 as weights
Change positive to non-negative
* reverseproxy: allow 0 as weighted round robin value
* test: add more wrr select test
---------
Co-authored-by: peanutduck <peanutduck@yahoo.com>
* caddyhttp: Add `MatchWithError` to replace SetVar hack
* Error in IP matchers on TLS handshake not complete
* Use MatchWithError everywhere possible
* Move implementations to MatchWithError versions
* Looser interface checking to allow fallback
* CEL factories can return RequestMatcherWithError
* Clarifying comment since it's subtle that an err is returned
* Return 425 Too Early status in IP matchers
* Keep AnyMatch signature the same for now
* Apparently Deprecated can't be all-uppercase to get IDE linting
* Linter
* metrics: move `metrics` up, outside `servers`
This change moves the metrics configuration from per-server level to a single config knob within the `http` app. Enabling `metrics` in any of the configured servers inside `http` enables metrics for all servers.
Fix#6604
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* normalize domain name
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
This fixes a regression in 2.8 where IP addresses
would be considered qualifying for public certs
by auto-HTTPS. The default issuers do not issue
IP certs at this time, so if no APs are explicitly
configured, we assign them to the internal
issuer. We have to add a couple lines of code because
CertMagic can no longer consider IPs as not
qualifying for public certs, since there are public CAs
that issue IP certs. This edge case is specific to Caddy's
auto-HTTPS.
Without this patch, Caddy will try using Let's Encrypt
or ZeroSSL's ACME endpoint to get IP certs, neither
of which support that.
* 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>
* caddyhttp: Escaping placeholders in CEL
* Simplify some of the test cases
* Implement vars and vars_regexp in CEL
* dupl lint is dumb
* Better consts for the placeholder CEL shortcut
* Bump CEL version, register a few extensions
* Refactor s390x test script for readability
* Add retries for s390x to smooth over flakiness
* Switch to `ph` for the CEL shortcut (match it in templates cause why not)
* caddy adapt for listen_protocols
* adapt listen_socket
* allow multiple listen sockets for port ranges and readd socket fd listen logic
* readd logic to start servers according to listener protocols
* gofmt
* adapt caddytest
* gosec
* fmt and rename listen to listenWithSocket
* fmt and rename listen to listenWithSocket
* more consistent error msg
* non unix listenReusableWithSocketFile
* remove unused func
* doc comment typo
* nonosec
* commit
* doc comments
* more doc comments
* comment was misleading, cardinality did not change
* addressesWithProtocols
* update test
* fd/ and fdgram/
* rm addr
* actually write...
* i guess we doin' "skip": now
* wrong var in placeholder
* wrong var in placeholder II
* update param name in comment
* dont save nil file pointers
* windows
* key -> parsedKey
* osx
* multiple default_bind with protocols
* check for h1 and h2 listener netw
* ci: update the linter action version
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* exclude rule `G115`; disable deprecated linter
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* reverseproxy: allow user to define source address
Closes#6503
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* reverse_proxy: caddyfile support for local_address
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* error: run `error` (msg) through replacer
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
* fix integration test
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
---------
Signed-off-by: Mohammed Al Sahaf <msaa1990@gmail.com>
Most of the errors that can be seen here are write errors due to clients
aborting the request from their side. Often seen ones include:
* writing: ... write: broken pipe
* writing: ... connection timed out
* writing: http2: stream closed
* writing: timeout...
* writing: h3 error...
Most of these errors are beyond of the control of caddy on the client side,
probably nothing can be done on the server side. It still warrants
researching when these errors occur very often, so a change in level from
error to warn is better here to not polute the logs with errors in the
normal case.
