Improve RFC links (#8612)

The `tools.ietf.org` site now redirects to `www.rfc-editor.org`.
* Update various URLs to the new site.
* Make links to `www.rfc-editor.org` consistent.

Signed-off-by: SuperQ <superq@gmail.com>
This commit is contained in:
Ben Kochie
2026-10-08 10:21:06 +02:00
committed by GitHub
parent 763bdb9b76
commit 311a9915da
60 changed files with 1006 additions and 215 deletions

View File

@@ -1,9 +1,9 @@
.\" Generated by Mmark Markdown Processer - mmark.miek.nl
.TH "COREDNS-TIMEOUTS" 7 "March 2026" "CoreDNS" "CoreDNS Plugins"
.TH "COREDNS-TIMEOUTS" 7 "October 2026" "CoreDNS" "CoreDNS Plugins"
.SH "NAME"
.PP
\fItimeouts\fP - allows you to configure the server read, write and idle timeouts for the TCP, TLS, DoH and DoQ (idle only) servers.
\fItimeouts\fP - allows you to configure the supported server read, write and idle timeouts for the TCP, TLS, DoH and DoQ servers, and the maximum number of queries served on a single TCP or TLS connection.
.SH "DESCRIPTION"
.PP
@@ -18,7 +18,8 @@ with such routers.
.PP
The \fItimeouts\fP "plugin" allows you to configure CoreDNS server read, write and
idle timeouts.
idle timeouts, and the maximum number of queries CoreDNS will serve on a single
TCP or TLS connection before closing it.
.SH "SYNTAX"
.PP
@@ -29,6 +30,7 @@ timeouts {
read DURATION
write DURATION
idle DURATION
maxtcpqueries MAXIMUM
}
.fi
@@ -36,9 +38,27 @@ timeouts {
.PP
For any timeouts that are not provided, default values are used which may vary
depending on the server type. At least one timeout must be specified otherwise
depending on the server type. At least one option must be specified otherwise
the entire timeouts block should be omitted.
.IP \(bu 4
\fB\fCmaxtcpqueries\fR sets the maximum number of queries served on a single TCP or
TLS connection before CoreDNS closes it. \fBMAXIMUM\fP must be a positive
integer, or \fB\fC-1\fR to allow an unlimited number of queries per connection
(the default). Long-lived connections that serve an unlimited number of
queries can cause uneven load distribution across CoreDNS replicas that sit
behind a connection-based load balancer, since new queries keep reusing the
same connection instead of establishing a new one. Setting a bound, e.g.
\fB\fCmaxtcpqueries 128\fR, forces clients to periodically reconnect, which allows
the load balancer to redistribute load.
.PP
The configured timeouts apply where the selected server transport supports
them. TCP, TLS and DoH servers use the read, write and idle timeouts. DoQ
servers use the read timeout to bound receiving a query on an opened QUIC
stream, and the idle timeout to bound idle QUIC connections.
.SH "EXAMPLES"
.PP
Start a DNS-over-TLS server that picks up incoming DNS-over-TLS queries on port
@@ -83,7 +103,8 @@ https://. {
.RE
.PP
Start a DNS-over-QUIC server that has the idle timeout set to two minutes.
Start a DNS-over-QUIC server that has a 10 second read timeout for receiving a
query on an opened stream and the idle timeout set to two minutes.
.PP
.RS
@@ -92,6 +113,7 @@ Start a DNS-over-QUIC server that has the idle timeout set to two minutes.
quic://.:853 {
tls cert.pem key.pem ca.pem
timeouts {
read 10s
idle 2m
}
forward . /etc/resolv.conf
@@ -119,3 +141,22 @@ configured. The timeouts are only applied to the TCP side of the server.
.fi
.RE
.PP
Start a standard TCP/UDP server that closes a TCP connection after it has
served 128 queries, to help spread load evenly across replicas sitting behind
a connection-based load balancer.
.PP
.RS
.nf
\&. {
timeouts {
maxtcpqueries 128
}
forward . /etc/resolv.conf
}
.fi
.RE