Wrap doPrefetch with a fresh metadata context to prevent concurrent
writes to the request-scoped metadata map during background prefetch.
Add a new integration test configuring a plugin chain, triggering
the issue seen here. Hammers concurrent queries while log reads
metadata fields repeatedly.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
Update coredns/caddy to a version where Dispenser.NextBlock()
checks Next() and stops at EOF. This ensures forward progress
and prevents an infinite loop when a block is missing a closing '}'
under certain conditions.
Added a regression test.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
Fix a panic in presubmit test when import statements are split into
>3 logical blocks (e.g., std, coredns, then third party in multiple
blocks). The computed block index could exceed the fixed array
bounds.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
Avoid transient EADDRINUSE and a previous negative WaitGroup panic:
- restart onto a different free port
- no Stop() around Restart()
- channel-coordinated Stop of the new instance
Fixes#7311
Signed-off-by: Syed Azeez <syedazeez337@gmail.com>
Enable protogetter in golangci config and update all protobuf field
access to use getter methods instead of direct field access.
Getter methods provide safer nil pointer handling and return
appropriate default values, following protobuf best practices.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
Enable intrange linter to enforce modern Go range syntax over
traditional for loops, by converting:
for i := 0; i < n; i++
to:
for i := range n
Adding type conversions where needed for compatibility
with existing uint64 parameters.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
Add conditional test skipping for bind and readme tests that rely on
Linux-specific loopback interface behavior. These tests reference
network configurations that may not exist on for e.g. macOS or other
platforms, causing spurious test failures.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
The rewrite plugin modifies DNS messages, affecting the request
size observed in the coredns_dns_request_size_bytes metric.
This change captures the original request size before any plugins
can modify it. It adds a functional options pattern to Report() to
pass this information while maintaining API compatibility.
Tests have been added to verify the fix prevents rewrite from
affecting the request size metrics.
Docs included.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
www.example.net is now behind akamai with various IP answered and a
chain of CNAME. Let's replace www.example.net by one of the root server
which answer a single IP and hopefully should remain this way.
Signed-off-by: Arthur Outhenin-Chalandre <arthur@cri.epita.fr>
* introduce new interface "dnsserver.Viewer", that allows a plugin implementing it to decide if a query should be routed into its server block.
* add new plugin "view", that uses the new interface to enable a user to define expression based conditions that must be met for a query to be routed to its server block.
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* write failures with ResponseReverter instead of letting server write them
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* fix comment
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* Metrics: expand coredns_dns_responses_total with plugin label
This adds (somewhat hacky?) code to add a plugin label to the
coredns_dns_responses_total metric. It's completely obvlious to the
plugin as we just check who called the *recorder.WriteMsg method. We use
runtime.Caller( 1 2 3) to get multiple levels of callers, this should be
deep enough, but it depends on the dns.ResponseWriter wrapping that's
occuring.
README.md of metrics updates and test added in test/metrics_test.go to
check for the label being set.
I went through the plugin to see what metrics could be removed, but
actually didn't find any, the plugin push out metrics that make sense.
Due to the path fiddling to figure out the plugin name I doubt this
works (out-of-the-box) for external plugins, but I haven't tested that.
Signed-off-by: Miek Gieben <miek@miek.nl>
* better comment
Signed-off-by: Miek Gieben <miek@miek.nl>
* Metrics: expand coredns_dns_responses_total with plugin label
This adds (somewhat hacky?) code to add a plugin label to the
coredns_dns_responses_total metric. It's completely obvlious to the
plugin as we just check who called the *recorder.WriteMsg method. We use
runtime.Caller( 1 2 3) to get multiple levels of callers, this should be
deep enough, but it depends on the dns.ResponseWriter wrapping that's
occuring.
README.md of metrics updates and test added in test/metrics_test.go to
check for the label being set.
I went through the plugin to see what metrics could be removed, but
actually didn't find any, the plugin push out metrics that make sense.
Due to the path fiddling to figure out the plugin name I doubt this
works (out-of-the-box) for external plugins, but I haven't tested that.
Signed-off-by: Miek Gieben <miek@miek.nl>
* Update core/dnsserver/server.go
Co-authored-by: dilyevsky <ilyevsky@gmail.com>
* Use [3]string
Signed-off-by: Miek Gieben <miek@miek.nl>
* imports
Signed-off-by: Miek Gieben <miek@miek.nl>
* remove dnstest changes
Signed-off-by: Miek Gieben <miek@miek.nl>
* revert
Signed-off-by: Miek Gieben <miek@miek.nl>
* Add some sleeps to make it less flaky
Signed-off-by: Miek Gieben <miek@miek.nl>
* Revert "Add some sleeps to make it less flaky"
This reverts commit b5c6655196.
* Remove forward when not needed
Signed-off-by: Miek Gieben <miek@miek.nl>
* remove newline
Signed-off-by: Miek Gieben <miek@miek.nl>
Co-authored-by: dilyevsky <ilyevsky@gmail.com>
* write cname answer to client even if target lookup is servfail
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* fix existing unit test expectations
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
Deflake by retrying and adding random port numbers. We try 3 times to
get an instance.
Also fix a bug where server.Stop() was called even if the server
creation failed - this was never hit due to t.Fatal() above it, but fix
that nontheless.
Signed-off-by: Miek Gieben <miek@miek.nl>
* share plugins among zones in the same server block
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* update caddy dep
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* simp code
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* copy ListenHosts and Debug from first config
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* copy tls configs from first config
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* add test to validate debug setting is replicated to all configs in block
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
* stop server
Signed-off-by: Chris O'Haver <cohaver@infoblox.com>
This was found by fuzzing.
We need to make this a fully qualified domain name to catch all errors
in dnsserver/register.go and not later when plugin.Normalize() is called again on these
strings, with the prime difference being that the domain name is fully
qualified. This was found by fuzzing where "ȶ" is deemed OK, but "ȶ." is
not (might be a bug in miekg/dns actually). But here we were checking ȶ,
which is OK, and later we barf in ȶ. leading to "index out of range".
Added a tests and check manually if it would crash with the current code
(yes), and fail with an error in this PR (yes).
Signed-off-by: Miek Gieben <miek@miek.nl>
* core: fix v4 non-octet reverse zones
This fixes the reverse zones handling. Add expanstion of the reverse
notation to all octet boundary subnets and add those to the config - just as if
they were directly typed in the config.
This takes inspiration from #4501, but that (even with DCO!!) seems to
be just using https://github.com/apparentlymart/go-cidr/ so use that
instead - I think a minor function is still needed that one is copied
from #4501.
Also sort the zones we are listing on startup - caught in this PR
because of the expanded zones being not listed next to each other.
This also removes the need for FilterFunc from the config, so this is
now gone as well, making the whole thing slightly more efficient.
Add couple of reverse unit tests and a e2e test that queries for the
correct (and incorrect) reverse zones and checks the reply.
Closes: #4501Fixes: #2779
Signed-off-by: Miek Gieben <miek@miek.nl>
* Add more test cases
Add test from origin bug report: #2779
Signed-off-by: Miek Gieben <miek@miek.nl>
* Rebase and fix conflicts
Signed-off-by: Miek Gieben <miek@miek.nl>
The following test failure for `TestZoneReload` can be observed
periodically:
`file_reload_test.go:58: Expected two RR in answer section got 2`
This failure can be consistently reproduced using the following command
(on my machine, at least):
`( cd test ; go test -v -race -run "TestZoneReload" ./... -count=500)`
test/file_reload_test.go:
Address a typo in a test failure message.
Sleep for double the file reload interval to avoid a rare
race condition between test code and the file plugin's reload code,
which is presumably a result of the time it takes to actually reload.
Signed-off-by: Stephen Greene <sgreene@redhat.com>
* plugin/transfer: only allow outgoing axfr over tcp
Return refused when the query comes in over udp.
No need to add a new test case as the current crop needed to be changed
to use TCP.
Fixes: #4450
Signed-off-by: Miek Gieben <miek@miek.nl>
* transfer tests: this needs tcp as well
Signed-off-by: Miek Gieben <miek@miek.nl>
