* plugin/pkg/proxy: add max_age for per-connection lifetime cap
Introduce a max_age setting on Transport that closes connections based
on creation time, independent of idle-timeout (expire).
Background: PR #7790 changed the connection pool from LIFO to FIFO for
source-port diversity. Under FIFO, every connection is cycled through
the pool and its used timestamp is refreshed continuously. When request
rate is high enough that pool_size / request_rate < expire, no
connection ever becomes idle and expire never fires. This prevents
CoreDNS from opening new connections to upstreams that scale out (e.g.
new Kubernetes pods behind a ClusterIP service with conntrack pinning).
max_age addresses this by enforcing an absolute upper bound on
connection lifetime regardless of activity:
- persistConn gains a created field set at dial time.
- Transport gains maxAge (default 0 = unlimited, preserving existing
behaviour).
- Dial(): rejects cached connections whose creation age exceeds max_age.
- cleanup(): when maxAge > 0, uses a linear scan over both idle-timeout
and max-age predicates; when maxAge == 0, preserves the original
binary-search path on used time (sorted by FIFO insertion order).
- Both hot paths pre-compute the deadline outside any inner loop to
avoid repeated time.Now() calls.
Tests added:
- TestMaxAgeExpireByCreation: connection with old created but fresh used
must be rejected even though idle-timeout would pass.
- TestMaxAgeFIFORotation: three FIFO-rotated connections (old created,
fresh used) must all be rejected, confirming that continuous rotation
cannot prevent max-age expiry.
Signed-off-by: cangming <cangming@cangming.app>
* plugin/forward: add max_age option
Expose Transport.SetMaxAge through the forward plugin so operators can
set an absolute upper bound on connection lifetime via the Corefile.
Usage:
forward . 1.2.3.4 {
max_age 30s
}
Default is 0 (unlimited), which preserves existing behaviour.
A positive value causes connections older than max_age to be closed and
re-dialled on the next request, ensuring CoreDNS reconnects to newly
scaled-out upstream pods even under sustained load where the idle
timeout (expire) would never fire.
If max_age is set, it must not be less than expire; the parser rejects
this combination at startup with a clear error message.
Signed-off-by: cangming <cangming@cangming.app>
---------
Signed-off-by: cangming <cangming@cangming.app>
* perf(proxy): use mutex-based connection pool
The proxy package (used for example by the forward plugin) utilized
an actor model where a single connManager goroutine managed
connection pooling via unbuffered channels (dial, yield, ret). This
design serialized all connection acquisition and release operations
through a single goroutine, creating a bottleneck under high
concurrency. This was observable as a performance degradation when
using a single upstream backend compared to multiple backends
(which sharded the bottleneck).
Changes:
- Removed dial, yield, and ret channels from the Transport struct.
- Removed the connManager goroutine's request processing loop.
- Implemented Dial() and Yield() using a sync.Mutex to protect the
connection slice, allowing for fast concurrent access without
context switching.
- Downgraded connManager to a simple background cleanup loop that
only handles connection expiration on a ticker.
- Updated plugin/pkg/proxy/connect.go to use direct method calls
instead of channel sends.
- Updated tests to reflect the removal of internal channels.
Benchmarks show that this change eliminates the single-backend
bottleneck. Now a single upstream backend performs on par with
multiple backends, and overall throughput is improved.
The implementation aligns with standard Go patterns for connection
pooling (e.g., net/http.Transport).
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
* fix: address PR review for persistent.go
- Named mutex field instead of embedding, to not expose
Lock() and Unlock()
- Move stop check outside of lock in Yield()
- Close() without a separate goroutine
- Change stop channel to struct
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
* fix: address code review feedback for conn pool
- Switch from LIFO to FIFO connection selection for source port
diversity, reducing DNS cache poisoning risk (RFC 5452).
- Remove "clear entire cache" optimization as it was LIFO-specific.
FIFO naturally iterates and skips expired connections.
- Remove all goroutines for closing connections; collect connections
while holding lock, close synchronously after releasing lock.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
* fix: remove unused error consts
No longer utilised after refactoring the channel based approach.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
* feat(forward): add max_idle_conns option
Add configurable connection pool limit for the forward plugin via
the max_idle_conns Corefile option.
Changes:
- Add SetMaxIdleConns to proxy
- Add maxIdleConns field to Forward struct
- Add max_idle_conns parsing in forward plugin setup
- Apply setting to each proxy during configuration
- Update forward plugin README with new option
By default the value is 0 (unbounded). When set, excess
connections returned to the pool are closed immediately
rather than cached.
Also add a yield related test.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
* chore(proxy): simple Dial by closing conns inline
Remove toClose slice collection to reduce complexity. Instead close
expired connections directly while iterating. Reduces complexity with
negligible lock-time impact.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
* chore: fewer explicit Unlock calls
Cleaner and less chance of forgetting to unlock on new possible
code paths.
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
---------
Signed-off-by: Ville Vesilehto <ville@vesilehto.fi>
* plugin/forward: Move Proxy into pkg/plugin/proxy, to allow forward.Proxy to be used outside of forward plugin.
Signed-off-by: Patrick Downey <patrick.downey@dioadconsulting.com>
