shaba/systemd
1
0
Fork 0
mirror of https://github.com/systemd/systemd.git synced 2024-12-26 03:22:00 +03:00
Code

resolve: do not derive query timeout from RTT

Browse Source 
DNS queries need timeout values to detect whether a DNS server is
unresponsive or, if the query is sent over UDP, whether a DNS message
was lost and has to be resent. The total time that it takes to answer a
query to arrive is t + RTT, where t is the maximum time that the DNS
server that is being queried needs to answer the query.

An authoritative server stores a copy of the zone that it serves in main
memory or secondary storage, so t is very small and therefore the time
that it takes to answer a query is almost entirely determined by the
RTT. Modern authoritative server software keeps its zones in main memory
and, for example, Knot DNS and NSD are able to answer in less than
100 µs [1]. So iterative resolvers continuously measure the RTT to
optimize their query timeouts and to resend queries more quickly if they
are lost.

systemd-resolved is a stub resolver: it forwards DNS queries to an
upstream resolver and waits for an answer. So the time that it takes for
systemd-resolved to answer a query is determined by the RTT and the time
that it takes the upstream resolver to answer the query.

It seems common for iterative resolver software to set a total timeout
for the query. Such total timeout subsumes the timeout of all queries
that the iterative has to make to answer a query. For example, BIND
seems to use a default timeout of 10 s.

At the moment systemd-resolved derives its query timeout entirely from
the RTT and does not consider the query timeout of the upstream
resolver. Therefore it often mistakenly degrades the feature set of its
upstream resolvers if it takes them longer than usual to answer a query.
It has been reported to be a considerable problem in practice, in
particular if DNSSEC=yes. So the query timeout systemd-resolved should
be derived from the timeout of the upstream resolved and the RTT to the
upstream resolver.

At the moment systemd-resolved measures the RTT as the time that it
takes the upstream resolver to answer a query. This clearly leads to
incorrect measurements. In order to correctly measure the RTT
systemd-resolved would have to measure RTT separately and continuously,
for example with a query with an empty question section or a query for
the SOA RR of the root zone so that the upstream resolver would be able
to answer to query without querying another server. However, this
requires significant changes to systemd-resolved. So it seems best to
postpone them until other issues have been addressed and to set the
resend timeout to a fixed value for now.

As mentioned, BIND seems to use a timeout of 10 s, so perhaps 12 s is a
reasonable value that also accounts for common RTT values. If we assume
that the we are going to retry, it could be less. So it should be enough
to set the resend timeout to DNS_TIMEOUT_MAX_USEC as
DNS_SERVER_FEATURE_RETRY_ATTEMPTS * DNS_TIMEOUT_MAX_USEC = 15 s.
However, this will not solve the incorrect feature set degradation and
should be seen as a temporary change until systemd-resolved does
probe the feature set of an upstream resolver independently from the
actual queries.

[1] https://www.knot-dns.cz/benchmark/
This commit is contained in:
Matthias-Christian Ott 2018-06-11 20:07:36 +02:00 committed by Lennart Poettering
parent 17ae278097
commit dbc4661a2c
3 changed files with 7 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
src/resolve/resolved-dns-server.c
View File

@ -15,10 +15,6 @@
#include "string-table.h"
#include "string-util.h"
/* After how much time to repeat classic DNS requests */
#define DNS_TIMEOUT_MIN_USEC (750 * USEC_PER_MSEC)
#define DNS_TIMEOUT_MAX_USEC (SD_RESOLVED_QUERY_TIMEOUT_USEC / DNS_TRANSACTION_ATTEMPTS_MAX)
/* The amount of time to wait before retrying with a full feature set */
#define DNS_SERVER_FEATURE_GRACE_PERIOD_MAX_USEC (6 * USEC_PER_HOUR)
#define DNS_SERVER_FEATURE_GRACE_PERIOD_MIN_USEC (5 * USEC_PER_MINUTE)
@ -265,7 +261,7 @@ static void dns_server_reset_counters(DnsServer *s) {
* incomplete. */
}
void dns_server_packet_received(DnsServer *s, int protocol, DnsServerFeatureLevel level, usec_t rtt, size_t size) {
void dns_server_packet_received(DnsServer *s, int protocol, DnsServerFeatureLevel level, size_t size) {
assert(s);
if (protocol == IPPROTO_UDP) {
@ -304,14 +300,6 @@ void dns_server_packet_received(DnsServer *s, int protocol, DnsServerFeatureLeve
this size. */
if (protocol == IPPROTO_UDP && s->received_udp_packet_max < size)
s->received_udp_packet_max = size;
if (s->max_rtt < rtt) {
s->max_rtt = rtt;
s->resend_timeout = CLAMP(s->max_rtt * 2, DNS_TIMEOUT_MIN_USEC, DNS_TIMEOUT_MAX_USEC);
} else if (s->resend_timeout > rtt)
/* If we received the packet faster than the resend_timeout, bias
* the resend_timeout back to the rtt. */
s->resend_timeout = CLAMP((2 * s->resend_timeout + rtt) / 3, DNS_TIMEOUT_MIN_USEC, DNS_TIMEOUT_MAX_USEC);
}
void dns_server_packet_lost(DnsServer *s, int protocol, DnsServerFeatureLevel level, usec_t usec) {
@ -328,11 +316,6 @@ void dns_server_packet_lost(DnsServer *s, int protocol, DnsServerFeatureLevel le
s->n_failed_tcp++;
}
}
if (s->resend_timeout > usec)
return;
s->resend_timeout = MIN(s->resend_timeout * 2, DNS_TIMEOUT_MAX_USEC);
}
void dns_server_packet_truncated(DnsServer *s, DnsServerFeatureLevel level) {
@ -859,9 +842,6 @@ void dns_server_flush_cache(DnsServer *s) {
void dns_server_reset_features(DnsServer *s) {
assert(s);
s->max_rtt = 0;
s->resend_timeout = DNS_TIMEOUT_MIN_USEC;
s->verified_feature_level = _DNS_SERVER_FEATURE_LEVEL_INVALID;
s->possible_feature_level = DNS_SERVER_FEATURE_LEVEL_BEST;

5
src/resolve/resolved-dns-server.h
View File

@ -67,9 +67,6 @@ struct DnsServer {
gnutls_datum_t tls_session_data;
#endif
usec_t resend_timeout;
usec_t max_rtt;
DnsServerFeatureLevel verified_feature_level;
DnsServerFeatureLevel possible_feature_level;
@ -112,7 +109,7 @@ DnsServer* dns_server_unref(DnsServer *s);
void dns_server_unlink(DnsServer *s);
void dns_server_move_back_and_unmark(DnsServer *s);
void dns_server_packet_received(DnsServer *s, int protocol, DnsServerFeatureLevel level, usec_t rtt, size_t size);
void dns_server_packet_received(DnsServer *s, int protocol, DnsServerFeatureLevel level, size_t size);
void dns_server_packet_lost(DnsServer *s, int protocol, DnsServerFeatureLevel level, usec_t usec);
void dns_server_packet_truncated(DnsServer *s, DnsServerFeatureLevel level);
void dns_server_packet_rrsig_missing(DnsServer *s, DnsServerFeatureLevel level);

8
src/resolve/resolved-dns-transaction.c
View File

@ -25,6 +25,9 @@
#define TRANSACTIONS_MAX 4096
#define TRANSACTION_TCP_TIMEOUT_USEC (10U*USEC_PER_SEC)
/* After how much time to repeat classic DNS requests */
#define DNS_TIMEOUT_USEC (SD_RESOLVED_QUERY_TIMEOUT_USEC / DNS_TRANSACTION_ATTEMPTS_MAX)
static void dns_transaction_reset_answer(DnsTransaction *t) {
assert(t);
@ -1137,7 +1140,7 @@ void dns_transaction_process_reply(DnsTransaction *t, DnsPacket *p) {
dns_server_packet_bad_opt(t->server, t->current_feature_level);
/* Report that we successfully received a packet */
dns_server_packet_received(t->server, p->ipproto, t->current_feature_level, ts - t->start_usec, p->size);
dns_server_packet_received(t->server, p->ipproto, t->current_feature_level, p->size);
}
/* See if we know things we didn't know before that indicate we better restart the lookup immediately. */
@ -1355,8 +1358,7 @@ static usec_t transaction_get_resend_timeout(DnsTransaction *t) {
if (t->stream)
return TRANSACTION_TCP_TIMEOUT_USEC;
assert(t->server);
return t->server->resend_timeout;
return DNS_TIMEOUT_USEC;
case DNS_PROTOCOL_MDNS:
assert(t->n_attempts > 0);