Fix DNS-over-HTTP(S) implementation issues that arise under heavy
query load. Optimize resource usage for :iscman:`named` instances
that accept queries over DNS-over-HTTP(S).
Previously, :iscman:`named` would process all incoming HTTP/2 data
at once, which could overwhelm the server, especially when dealing
with clients that send requests but don't wait for responses. That
has been fixed. Now, :iscman:`named` handles HTTP/2 data in smaller
chunks and throttles reading until the remote side reads the
response data. It also throttles clients that send too many requests
at once.
Additionally, :iscman:`named` now carefully processes data sent by
some clients, which can be considered "flooding." It logs these
clients and drops connections from them.
:gl:`#4795`
In some cases, :iscman:`named` could leave DNS-over-HTTP(S)
connections in the `CLOSE_WAIT` state indefinitely. That also has
been fixed. ISC would like to thank JF Billaud for thoroughly
investigating the issue and verifying the fix.
:gl:`#5083`
See https://gitlab.isc.org/isc-projects/bind9/-/issues/4795
Closes https://gitlab.isc.org/isc-projects/bind9/-/issues/5083
Merge branch 'artem-improve-doh-resource-usage' into 'v9.21.4-release'
See merge request isc-private/bind9!732
We started using isc_nm_bad_request() more actively throughout
codebase. In the case of HTTP/2 it can lead to a large count of
useless "Bad Request" messages in the BIND log, as often we attempt to
send such request over effectively finished HTTP/2 sessions.
This commit fixes that.
A call to isc_nm_read_stop() would always stop reading timer even in
manual timer control mode which was added with StreamDNS in mind. That
looks like an omission that happened due to how timers are controlled
in StreamDNS where we always stop the timer before pausing reading
anyway (see streamdns_on_complete_dnsmessage()). That would not work
well for HTTP, though, where we might want pause reading without
stopping the timer in the case we want to split incoming data into
multiple chunks to be processed independently.
I suppose that it happened due to NM refactoring in the middle of
StreamDNS development (at the time isc_nm_cancelread() and
isc_nm_pauseread() were removed), as the StreamDNS code seems to be
written as if timers are not stoping during a call to
isc_nm_read_stop().
This commit introduces manual read timer control as used by StreamDNS
and its underlying transports. Before that, DoH code would rely on the
timer control provided by TCP, which would reset the timer any time
some data arrived. Now, the timer is restarted only when a full DNS
message is processed in line with other DNS transports.
That change is required because we should not stop the timer when
reading from the network is paused due to throttling. We need a way to
drop timed-out clients, particularly those who refuse to read the data
we send.
This commit adds logic to make code better protected against clients
that send valid HTTP/2 data that is useless from a DNS server
perspective.
Firstly, it adds logic that protects against clients who send too
little useful (=DNS) data. We achieve that by adding a check that
eventually detects such clients with a nonfavorable useful to
processed data ratio after the initial grace period. The grace period
is limited to processing 128 KiB of data, which should be enough for
sending the largest possible DNS message in a GET request and then
some. This is the main safety belt that would detect even flooding
clients that initially behave well in order to fool the checks server.
Secondly, in addition to the above, we introduce additional checks to
detect outright misbehaving clients earlier:
The code will treat clients that open too many streams (50) without
sending any data for processing as flooding ones; The clients that
managed to send 1.5 KiB of data without opening a single stream or
submitting at least some DNS data will be treated as flooding ones.
Of course, the behaviour described above is nothing else but
heuristical checks, so they can never be perfect. At the same time,
they should be reasonable enough not to drop any valid clients,
realatively easy to implement, and have negligible computational
overhead.
Initially, our DNS-over-HTTP(S) implementation would try to process as
much incoming data from the network as possible. However, that might
be undesirable as we might create too many streams (each effectively
backed by a ns_client_t object). That is too forgiving as it might
overwhelm the server and trash its memory allocator, causing high CPU
and memory usage.
Instead of doing that, we resort to processing incoming data using a
chunk-by-chunk processing strategy. That is, we split data into small
chunks (currently 256 bytes) and process each of them
asynchronously. However, we can process more than one chunk at
once (up to 4 currently), given that the number of HTTP/2 streams has
not increased while processing a chunk.
That alone is not enough, though. In addition to the above, we should
limit the number of active streams: these streams for which we have
received a request and started processing it (the ones for which a
read callback was called), as it is perfectly fine to have more opened
streams than active ones. In the case we have reached or surpassed the
limit of active streams, we stop reading AND processing the data from
the remote peer. The number of active streams is effectively decreased
only when responses associated with the active streams are sent to the
remote peer.
Overall, this strategy is very similar to the one used for other
stream-based DNS transports like TCP and TLS.
When answering queries, don't add data to the additional section if the answer has more than 13 names in the RDATA. This limits the number of lookups into the database(s) during a single client query, reducing query processing load.
See isc-projects/bind9#5034
Merge branch '5034-security-limit-additional' into 'v9.21.4-release'
See merge request isc-private/bind9!750
Limit the number of records appended to ADDITIONAL section to the names
that have less than 14 records in the RDATA. This limits the number
of the lookups into the database(s) during single client query.
Also don't append any additional data to ANY queries. The answer to ANY
is already big enough.
Instead of running the whole resolver/ns4 server with -T noaa flag,
use it only for the part where it is actually needed. The -T noaa
could interfere with other parts of the test because the answers don't
have the authoritative-answer bit set, and we could have false
positives (or false negatives) in the test because the authoritative
server doesn't follow the DNS protocol for all the tests in the resolver
system test.
All the database implementations share the same names for the methods
implementing the database. That has some advantages like knowing what
to expect, but it turns out that any time such method shows up in any
kind of tracing - be it perf record, backtrace or anything else that
uses symbol names, it is very hard to distinguish whether the find()
belongs to qpcache, qpzone, builtin or sdlz implementation.
Make at least the names for qpzone and qpcache unique.
A bug in the qpzone database could trigger a crash when querying for a deleted name, or a newly-added empty non-terminal name, in an NSEC3-signed zone. This has been fixed.
Closes#5108
Merge branch '5108-nsec3-empty-node' into 'main'
See merge request isc-projects/bind9!9928
there was a database bug in which dns_db_find() could get a partial
match for the query name, but still set foundname to match the full
query name. this triggered an assertion when query_addwildcardproof()
assumed that foundname would be shorter.
the database bug has been fixed, but in case it happens again, we
can just copy the name instead of splitting it. we will also log a
warning that the closest-encloser name was invalid.
when adding a new NSEC3 record, dns_nsec3_addnsec3() uses a
dbiterator to seek to the newly created node and then find its
predecessor. dbiterators in the qpzone use snapshots, so changes
to the database are not reflected in an already-existing iterator.
consequently, when we add a new node, we have to create a new iterator
before we can seek to it.
this test adds a record with empty non-terminal nodes above it. this
has also been observed to trigger the crash in NSEC3 zones.
NOTE: the test currently fails, because while there is no crash, the
query results are not as expected. when we add a node below an ENT,
receive_secure_serial() gets DNS_R_PARTIALMATCH, and the signed
zone is never updated. this is not a regression from fixing the
crash bug; it's a separate inline-signing bug.
test that there's no crash when querying for a newly-deleted node.
(incidentally also renamed ns3/named.conf.in to ns3/named1.conf.in,
because named2.conf.in does exist, and they should match.)
when a requested name is found in the QP trie during a lookup, but its
records have been marked as nonexistent by a previous deletion, then
it's treated as a partial match, but the foundname could be left
pointing to the original qname rather than the parent. this could
lead to an assertion failure in query_findclosestnsec3().
Commit b121f02eac342ee285b6ab1292a0136448a91ee0 renamed the top-level
"primaries" block in bin/named/config.c to "remote-servers". This
configuration block lists the primary servers used for an IANA root zone
mirror when no primary servers are explicitly specified for it in the
configuration. However, the relevant part of the named_zone_configure()
function only looks for a top-level "primaries" block and not for any of
its synonyms. As a result, configuring an IANA root zone mirror with
just:
zone "." {
type mirror;
};
now results in a cryptic fatal error on startup:
loading configuration: not found
exiting (due to fatal error)
Fix by using the correct top-level block name in named_zone_configure().
Response policy zones (RPZ) and catalog zones were not working correctly if they had an $INCLUDE statement defined. This has been fixed.
Closes#5111
Merge branch '5111-includes-disable-rpz-and-catz-fix' into 'main'
See merge request isc-projects/bind9!9930
The code in zone_startload() disables RPZ and CATZ for a zone if
dns_master_loadfile() returns anything other than ISC_R_SUCCESS,
which makes sense, but it's an error because zone_startload() can
also return DNS_R_SEENINCLUDE upon success when the zone had an
$INCLUDE statement.
Add performance tests of DoH using the GET protocol to nightly pipelines.
Merge branch 'nicki/ci-shotgun-doh-get' into 'main'
See merge request isc-projects/bind9!9926
Adjust number of zones down to 23 to match those present when testing in FIPS mode.
Closes#5097
Merge branch '5097-checking-startup-notify-rate-limit-fails-on-ol-8-fips' into 'main'
See merge request isc-projects/bind9!9919
Debug messages logging network traffic now include information about both sides of each communication channel rather than just one of them.
Closes#4345
Merge branch '4345-log-both-from-and-to-socket-in-debug-messages' into 'main'
See merge request isc-projects/bind9!8349
Ensure the log prefixes passed to the dns_message_logpacketfrom()
function by its callers do not include the word "from" as the latter is
now emitted by the logfmtpacket() helper function.
Ensure the log prefixes passed to the dns_message_logpacketfromto()
function by its callers do not include the words "from" or "to" as those
are now emitted by the logfmtpacket() helper function.
Move dns_dispentry_getlocaladdress() calls around so that they are not
only invoked when dnstap support is compiled in. This function calls
isc_nmhandle_localaddr(), which may issue a system call, but only if the
ISC_SOCKET_DETAILS preprocessor macro is set at compile time.
Pass the value extracted by dns_dispentry_getlocaladdress() to
dns_message_logpacketfromto() so that it gets logged, adding useful
information to the relevant debug messages.
Since dns_message_logpacket() only takes a single socket address as a
parameter (and it is always the sending socket's address), rename it to
dns_message_logpacketfrom() so that its name better conveys its purpose
and so that the difference in purpose between this function and
dns_message_logpacketfromto() becomes more apparent.
Since dns_message_logfmtpacket() needs to be provided with both "from"
and "to" socket addresses, rename it to dns_message_logpacketfromto() so
that its name better conveys its purpose. Clean up the code comments
for that function.
Change the function prototype for dns_message_logfmtpacket() so that it
takes two isc_sockaddr_t parameters: one for the sending side and
another one for the receiving side. This enables debug messages to be
more precise.
Also adjust the function prototype for logfmtpacket() accordingly.
Unlike dns_message_logfmtpacket(), this function must not require both
'from' and 'to' parameters to be non-NULL as it is still going to be
used by dns_message_logpacket(), which only provides a single socket
address. Adjust its log format to handle both of these cases properly.
Adjust both dns_message_logfmtpacket() call sites accordingly, without
actually providing the second socket address yet. (This causes the
revised REQUIRE() assertion in dns_message_logfmtpacket() to fail; the
issue will be addressed in a separate commit.)
Both existing callers of the dns_message_logfmtpacket() function set the
argument passed as 'style' to &dns_master_style_comment. To simplify
these call sites, drop the 'style' parameter from the prototype for
dns_message_logfmtpacket() and use a fixed value of
&dns_master_style_comment in the function's body instead.
All callers of the logfmtpacket() helper function require the argument
passed as 'address' to be non-NULL. Meanwhile, the 'newline' and
'space' local variables in logfmtpacket() are only set to values
different than their initial values if the 'address' parameter is NULL.
Replace the 'newline' and 'space' local variables in logfmtpacket() with
fixed strings to improve code readability.
Enable extracting the exact address/port that a local wildcard/TCP socket is bound to, improving the accuracy of dnstap logging and providing more information in debug logs produced by system tests. Since this requires issuing an extra system call on some hot paths, this new feature is only enabled when the ``ISC_SOCKET_DETAILS`` preprocessor macro is set at compile time.
Closes#4344
Merge branch '4344-enable-extraction-of-exact-local-socket-addresses' into 'main'
See merge request isc-projects/bind9!8348
Extracting the exact address that each wildcard/TCP socket is bound to
locally requires issuing the getsockname() system call, which libuv
exposes via its uv_*_getsockname() functions. This is only required for
detailed logging and comes at a noticeable performance cost, so it
should not happen by default. However, it is useful for debugging
certain problems (e.g. cryptic system test failures), so a convenient
way of enabling that behavior should exist.
Update isc_nmhandle_localaddr() so that it calls uv_*_getsockname() when
the ISC_SOCKET_DETAILS preprocessor macro is set at compile time.
Ensure proper handling of sockets that wrap other sockets.
Set the new ISC_SOCKET_DETAILS macro by default when --enable-developer
is passed to ./configure. This enables detailed logging in the system
tests run in GitLab CI without affecting performance in non-development
BIND 9 builds.
Note that setting the ISC_SOCKET_DETAILS preprocessor macro at compile
time enables all callers of isc_nmhandle_localaddr() to extract the
exact address of a given local socket, which results e.g. in dnstap
captures containing more accurate information.
Mention the new preprocessor macro in the section of the ARM that
discusses why exact socket addresses may not be logged by default.
This MR is a prerequisite for !8348.
It intentionally does not have a changelog entry associated with it, to
prevent making a false impression of improving connection reuse **for
existing code**. It will only make a difference once !8348 gets merged
(and even then, only if the new `ISC_SOCKET_DETAILS` macro will be set
during build). That's because `isc_nmhandle_localaddr()` currently
simply returns `handle->local` and its return value will only be set to
the actual address the socket is bound to with !8348 in place.
Note that `dns_dispatch_gettcp()` is currently only used by the
`dns_request` API, so this MR's potential for introducing new breakage
is relatively low.
Closes#4693
Merge branch '4693-improve-reuse-of-outgoing-tcp-connections' into 'main'
See merge request isc-projects/bind9!8972
The dns_dispatch_gettcp() function is used for finding an existing TCP
connection that can be reused for sending a query from a specified local
address to a specified remote address. The logic for matching the
provided <local address, remote address> tuple to one of the existing
TCP connections is implemented in the dispatch_match() function:
- if the examined TCP connection already has a libuv handle assigned,
it means the connection has already been established; therefore,
compare the provided <local address, remote address> tuple against
the corresponding address tuple for the libuv handle associated with
the connection,
- if the examined TCP connection does not yet have a libuv handle
assigned, it means the connection has not yet been established;
therefore, compare the provided <local address, remote address>
tuple against the corresponding address tuple that the TCP
connection was originally created for.
This logic limits TCP connection reuse potential as the libuv handle
assigned to an existing dispatch object may have a more specific local
<address, port> tuple associated with it than the local <address, port>
tuple that the dispatch object was originally created for. That's
because the local address for outgoing connections can be set to a
wildcard <address, port> tuple (indicating that the caller does not care
what source <address, port> tuple will be used for establishing the
connection, thereby delegating the task of picking it to the operating
system) and then get "upgraded" to a specific <address, port> tuple when
the socket is bound (and a libuv handle gets associated with it). When
another dns_dispatch_gettcp() caller then tries to look for an existing
TCP connection to the same peer and passes a wildcard address in the
local part of the tuple, the function will not match that request to a
previously-established TCP connection (unless isc_nmhandle_localaddr()
returns a wildcard address as well).
Simplify dispatch_match() so that the libuv handle associated with an
existing dispatch object is not examined for the purpose of matching it
to the provided <local address, remote address> tuple; instead, always
examine the <local address, remote address> tuple that the dispatch
object was originally created for. This enables reuse of TCP
connections created without providing a specific local socket address
while still preventing other connections (created for a specific local
socket address) from being inadvertently shared.
This change ensures that SNI extension is used in outgoing connections over TLS (e.g. for DoT and DoH) when applicable.
Closes#5099
Merge branch 'artem-outgoing-tls-sni-support' into 'main'
See merge request isc-projects/bind9!9923
This commit ensures that BIND enables TLS SNI support for outgoing DoT
connections (when possible) in order to improve compatibility with
other DNS server software.
This commit adds support for setting SNI hostnames in outgoing
connections over TLS.
Most of the changes are related to either adapting the code to accept
and extra argument in *connect() functions and a couple of changes to
the TLS Stream to actually make use of the new SNI hostname
information.
ThreadSanitizer has trouble understanding that gluelist->glue is
constant after it is assigned to the slabheader with cmpxchg. Help
ThreadSanitizer to understand the code by using CMM_STORE_SHARED and
CMM_LOAD_SHARED on gluelist->glue.
Merge branch 'ondrej/hint-tsan-in-addglue' into 'main'
See merge request isc-projects/bind9!9929
Previously, we had an ISC_CONSTEXPR macro that was expanded to either
`constexpr` or `static const`, depending on compiler support. To make
the code cleaner, move `constexpr` support detection to Autoconf; if
`constexpr` support is missing from the compiler, define `constexpr` as
`static const` in config.h.
Merge branch 'ondrej/rename-ISC_CONSTEXPR-to-constexpr-shim' into 'main'
See merge request isc-projects/bind9!9924
Previously, we had an ISC_CONSTEXPR macro that was expanded to either
`constexpr` or `static const`, depending on compiler support. To make
the code cleaner, move `constexpr` support detection to Autoconf; if
`constexpr` support is missing from the compiler, define `constexpr` as
`static const` in config.h.
