in the "refactor tcpquota and pipeline refs" commit, the counting
of active interfaces was tightened in such a way that named could
fail to listen on an interface if there were more interfaces than
tcp-clients. when checking the quota to start accepting on an
interface, if the number of active clients was above zero, then
it was presumed that some other client was able to handle accepting
new connections. this, however, ignored the fact that the current client
could be included in that count, so if the quota was already exceeded
before all the interfaces were listening, some interfaces would never
listen.
we now check whether the current client has been marked active; if so,
then the number of active clients on the interface must be greater
than 1, not 0.
(cherry picked from commit 0b4e2cd4c3192ba88569dd344f542a8cc43742b5)
(cherry picked from commit d01023aaac)
- if the TCP quota has been exceeded but there are no clients listening
for new connections on the interface, we can now force attachment to the
quota using isc_quota_force(), instead of carrying on with the quota not
attached.
- the TCP client quota is now referenced via a reference-counted
'ns_tcpconn' object, one of which is created whenever a client begins
listening for new connections, and attached to by members of that
client's pipeline group. when the last reference to the tcpconn
object is detached, it is freed and the TCP quota slot is released.
- reduce code duplication by adding mark_tcp_active() function.
- convert counters to atomic.
(cherry picked from commit 7e8222378ca24f1302a0c1c638565050ab04681b)
(cherry picked from commit 4939451275722bfda490ea86ca13e84f6bc71e46)
(cherry picked from commit 13f7c918b8)
- ensure that tcpactive is cleaned up correctly when accept() fails.
- set 'client->tcpattached' when the client is attached to the tcpquota.
carry this value on to new clients sharing the same pipeline group.
don't call isc_quota_detach() on the tcpquota unless tcpattached is
set. this way clients that were allowed to accept TCP connections
despite being over quota (and therefore, were never attached to the
quota) will not inadvertently detach from it and mess up the
accounting.
- simplify the code for tcpquota disconnection by using a new function
tcpquota_disconnect().
- before deciding whether to reject a new connection due to quota
exhaustion, check to see whether there are at least two active
clients. previously, this was "at least one", but that could be
insufficient if there was one other client in READING state (waiting
for messages on an open connection) but none in READY (listening
for new connections).
- before deciding whether a TCP client object can to go inactive, we
must ensure there are enough other clients to maintain service
afterward -- both accepting new connections and reading/processing new
queries. A TCP client can't shut down unless at least one
client is accepting new connections and (in the case of pipelined
clients) at least one additional client is waiting to read.
(cherry picked from commit c7394738b2445c16f728a88394864dd61baad900)
(cherry picked from commit e965d5f11d3d0f6d59704e614fceca2093cb1856)
(cherry picked from commit 87d4311614)
Track pipeline groups using a shared reference counter
instead of a linked list.
(cherry picked from commit 513afd33eb17d5dc41a3f0d2d38204ef8c5f6f91)
(cherry picked from commit 9446629b73)
the TCP client quota could still be ineffective under some
circumstances. this change:
- improves quota accounting to ensure that TCP clients are
properly limited, while still guaranteeing that at least one client
is always available to serve TCP connections on each interface.
- uses more descriptive names and removes one (ntcptarget) that
was no longer needed
- adds comments
(cherry picked from commit 924651f1d5e605cd186d03f4f7340bcc54d77cc2)
(cherry picked from commit 55a7a458e3)
tcp-clients settings could be exceeded in some cases by
creating more and more active TCP clients that are over
the set quota limit, which in the end could lead to a
DoS attack by e.g. exhaustion of file descriptors.
If TCP client we're closing went over the quota (so it's
not attached to a quota) mark it as mortal - so that it
will be destroyed and not set up to listen for new
connections - unless it's the last client for a specific
interface.
(cherry picked from commit f97131d21b97381cef72b971b157345c1f9b4115)
(cherry picked from commit 9689ffc485)
Key IDs may accidentally match dig output that is not the key ID (for
example the RRSIG inception or expiration time, the query ID, ...).
Search for key ID + signer name should prevent that, as that is what
only should occur in the RRSIG record, and signer name always follows
the key ID.
(cherry picked from commit 83473b9758)
Remove sleep calls from test, rely on wait_for_log(). Make
wait_for_log() and dnssec_loadkeys_on() fail the test if the
appropriate log line is not found.
Slightly adjust the echo_i() lines to print only the key ID (not the
key name).
(cherry picked from commit 67f0635f3c)
One second may not be enough for an NSEC3 chain change triggered by an
UPDATE message to complete. Wait up to 10 seconds when checking whether
a given NSEC3 chain change is complete in the "nsupdate" system test.
(cherry picked from commit f8746cddbc)
In the "nsupdate" system test, do not sleep before checking results of
changes which are expected to be processed synchronously, i.e. before
nsupdate returns.
(cherry picked from commit 1c8e5ea333)
Make bin/tests/system/ifconfig.bat also configure addresses ending with
9 and 10, so that the script is in sync with its Unix counterpart.
Update comments listing the interfaces created by ifconfig.{bat,sh} so
that they do not include addresses whose last octet is zero (since an
address like 10.53.1.0/24 is not a valid host address and thus the
aforementioned scripts do not even attempt configuring them).
(cherry picked from commit b6c1cdfffe)
On Windows, the bin/tests/system/dnssec/signer/example.db.signed file
contains carriage return characters at the end of each line. Remove
them before passing the aforementioned file to the awk script extracting
key IDs so that the latter can work properly.
(cherry picked from commit e4280ed9f5)
As signals are currently not handled by named on Windows, instances
terminated using signals are not able to perform a clean shutdown, which
involves e.g. removing the lock file. Thus, waiting for a given
instance's lock file to be removed beforing assuming it is shut down
is pointless on Windows, so do not even attempt it.
(cherry picked from commit 761ba4514f)
5213. [bug] win32: Eliminated a race which allowed named.exe running
as a service to be killed prematurely during shutdown.
[GL #978]
(cherry picked from commit e7332343ed)
When a Windows service receives a request to stop, it should not set its
state to SERVICE_STOPPED until it is completely shut down as doing that
allows the operating system to kill that service prematurely, which in
the case of named may e.g. prevent the PID file and/or the lock file
from being cleaned up.
Set service state to SERVICE_STOP_PENDING when named begins its shutdown
and only report the SERVICE_STOPPED state immediately before exiting.
(cherry picked from commit 964749dfdb)
This tests both the cases when the DLV trust anchor is of an
unsupported or disabled algorithm, as well as if the DLV zone
contains a key with an unsupported or disabled algorithm.
(cherry picked from commit 3b7c849a3f)
The following changes were needed:
* Remove dnskey-sig-validity option (added in 9.12)
* Replace rndccmd, dig_with_opts with export variables
* Remove tests for CDNSKEY and CDS (in 9.11 always signed with ZSK)
Tinderbox UserandOndřej Surý
Ondřej Surý
Evan HuntandOndřej Surý
Michał KępieńandOndřej Surý
Witold KręcickiandOndřej Surý
Matthijs Mekking
Mark Andrews