If keyspace notifications are used at all, the respective objects and
threads must be created during startup. This requires at least some
keyspace to be configured. To support usage without any keyspaces
initially (and add them during runtime), add a special case (set
keyspace to -1).
Convert all keyspace variables to signed ints. Ignore negative keyspace
numbers where appropriate. Support Redis endpoint addresses without
database number.
Fixes#1902
Change-Id: I45a3c87bc515f9b14e64ec1ec0906dde271b5f8d
HTTP/WS init creates worker threads, which would be terminated by a fork
to background. Reverse the order.
Closes#1896
Possibly also relevant to #1895
Change-Id: I30b61e07ad3bed41b6b241e8943ed479277c1474
This causes a termination with extreme prejudice. Useful for debugging
memory leak issues as it bypasses shutdown cleanup.
Change-Id: Ic39c5ad4aa5a4f9dc28274732bb736956a47e5bc
Mark those SDPs which trigger processing
of music-on-hold.
Is defined by the config option:
`moh-attr-name`. If not defined then
is not used.
Will be useful for cases when it's important
for remote components to detect ongoing hold.
Change-Id: I0f2a284955dbc045b51920174f1f4fbefcb93907
Introduce `moh-max-repeats` configuration option,
which controls the `rtpe_config.moh_max_repeats`
used when initializing media player options
for the music-on-hold functionality.
By default is always set to 999,
if not defined otherwise.
Change-Id: I247dc532a6871934e44f9c4002313363356937fb
Make `moh-max-duration` available for both
music-on-hold functionality as well as
for the media player.
For that to work, do the following:
- keep `moh-max-duration` config option only for MoH,
if not set (so 0) by default is 1800000ms (half an hour)
- for the play media functionality introduce flag option
`repeat-duration`, by default is disabled
Policy changes:
- duration counter can be used in common with repeats
counter, but then takes a precedence over it.
Hence if first a duration is underflown, then EOF triggered.
Otherwise if the duration counter is still positive, but
repeats are negative, then do EOF based on repeats.
- the repeats counter will always count down during each
iteration, even when used together with the duration counter
For MoH to survive, the repeats counter is simple set to 999
to let the duration counter always win over repeats one
- MoH cannot take duration disabled, since otherwise
would make no sense for it. Hence always takes internally
defined value 1800000ms (half an hour) if not defined
by the configuration option
Backwards compatibility:
- is kept in regards of repeats counter
- is kept in regards of the play media functionality
Change-Id: I48ff3c17c9bed31f80c3106b275b703a9ccb4b26
Add duration limit for the moh functionality.
This is required due to a possibility to get
monologues and hence packet streams stuck
forever and sending audio to recipients
which already don't exist.
A configuration option that controls that:
`moh-max-duration` - gets a value in milliseconds.
By default is set to 1800000 (half an hour).
Change-Id: Id50a0a10ce5b52b3876a3122fb16a71accec90ff
When re-connecting to the remote redis server
try to re-resolve if the redist hostname
was an FQDN and not IP address.
Change-Id: Ie80e1d1a1ea76811c54123201ad4fe8cb64fc748
Make it so that options only need to be managed in one place, and all
code operating on the list of them is streamlined
Change-Id: Ia53b15910ec8973635e61cad1b43eff8d536a577
Add support for simplified G.107 formula, with math changed to
fixed-point. Retain legacy formula as an option for backwards
compatibility.
Change-Id: I999fc7de7be1407876c201c251538cea72b04008
Use provided macros for initrialise str objects everywhere instead of
manually setting .s and .len. This allows for flexibility to extend str
objects.
Change-Id: I5ebca0bc82b348301c2c8e94b974c7acbf249b82
Instead of keeping one global hash/list of all received ICE fragments,
indexed by string call ID and from-tag, move the hash/list of ICE
fragments into the call object, now indxed only by the from-tag.
This requires a call object to exist before an ICE fragment can be
stored. Change the order of operations to always create a call object
first, then parse the SDP, then check for ICE fragment processing. The
determining factor now is only whether the monologue exists.
Duplicate the raw SDP into the call's memory arena instead of onto the
heap.
ICE fragments now don't need to store the pre-parsed SDP any more, and
the trickle ICE handling code is guaranteed to always have the fully
parsed SDP information.
Strings for fragment indexes can now be allocated from the call's memory
arena.
Obsolete the extra thread used to delete old ICE fragments. This is now
done in-line with regular call timeout processing.
Change-Id: I7acbc4c52666c4cdf1c02324bf33cf0bfcb4f0d0
Use a BIO WRITE callback instead of BIO_read'ing from the BIO after each
operation. This is a more direct way to intercept data that needs to be
sent out.
Implement MTU-related BIO callbacks.
Deduct the assumed IP MTU overhead from the configured MTU during
startup.
Unlike the previous code, this does not necessarily send DTLS from the
same socket that received a message, nor to the same address that sent
one, and instead always uses the selected_sfd and ->endpoint. This may
or may not be a regression.
Closes#1806
Change-Id: I4d4456df3f378d00782cbfa64afdb2a038217e6c
Obsolete str_init(), rename STR_INIT() to just STR(), and replace all
instances of str_init() with STR().
no-op
Change-Id: I981529063ad2ea26089add467f7a84b638dbf423
Make sure we create the fallback buffer pool if the kernel module isn't
available but userspace operation is permitted.
Change-Id: I177c36b9a17c18f39e6c54ffb414c13fbb9c15ba
Revert of 128e84e and implement for io_uring poller
Turns out that this function is necessary to prevent out-of-order writes
to a streambuf once the connection is open.
Use a simple array to keep track of blocked fds under uring.
Change-Id: I4af2a64071030fd4892dde88547705230aec59fd
Use relaxed-order atomic ops to access config variables that can be set
during run time.
Rename config_lock to keyspaces_lock as the keyspaces list remains the
only object that it protects.
All other config variables aren't settable at runtime and so don't
require locks or atomic ops.
Change-Id: I7e01a34a4818a558d5648ae27a86f9880a95d050
The requirement to continuously update stats from the kernel has been
eliminates. Remove all relevant code.
Change-Id: Idc8093b64290b89ff3d015f3e61e4bbe25df6e23
Rename the "noop" method to "init" and split up the function to open the
table into two parts. Open the control file first, then do the mmap(),
then init other things, then finally call the init method.
Change-Id: I832f6e90fbec4375e3e19ed6e72d7cce78488e9e
Switch all memory buffers used for RTP I/O from generic stack or heap
allocated memory to the bufferpool implementation. Use a per-thread
bufferpool to minimise lock contention.
This commit is just a one-for-one swap and doesn't use the bufferpool's
reference counting semantics yet.
Change-Id: I9cba4ec97bd0afcd374bf6c0be2b608a46e73e57
Support is desired for Kamailio/Rtpengine traffic via UDP.
Adding homer-disable-rtcp-stats and homer-enable-ng config params to
separately control sending to Homer each traffic type. By default rtcp
is `on` when homer parameter is configured. NG is by default disabled.
closes#1802
Change-Id: Ib68fb133cffc5d8945f9b6bf60bab3e80fab9630
The poller-per-thread feature was broken with a division by zero. Take
the opportunity to rework it and eliminate the poller_map object. Use a
simple array of pollers for media sockets, plus one global poller for
control sockets. In the regular case only one poller is created and
everything points to that poller. In the poller-per-thread case, one
poller per thread is created, plus one poller (also with its own single
thread) for control connections. All control sockets use the single
control poller, while all media sockets get assigned one poller from the
pool in a round-robin fashion.
closes#1801
Change-Id: Iae91a3e10b7206455c6df33b1a472254c700ce21
poller_new used to have the side effect of initialising rtpe_now.
Restore this side effect explicitly in the startup code, so that timers
are launched with the correst start time instead of zero.
Change-Id: I590061a9665b52c4aed00c06dc330d2a226c73d3
For the use case with a separate base chain, we want to preempt other
firewall rules by default and insert our immediate jump rule at the
beginning, rather than at the end. Add an option to provide the other
behaviour.
Change-Id: I16171f7c028c89b94823ecc99387771be3ba9443
Convert each listener entity into a list.
Support a list of values for each option so that multiple
ports/addresses can be listed.
Keep previous behaviour unchanged: If ANY address is given, open
listeners for IPv4 and IPv6.
Change-Id: Ic54f28d1262f60d5e5c9d824a95e7c33ebc2aba9
All functions that create listener objects take a read-only endpoint, so
make them as const.
Remove the extra TOS argument to make all signatures the same.
Change-Id: I722c7665b192476d90dbf0ece200d0bfd34cb9eb
Forking to background kills already created threads, so forking must
happen first and thread creation after.
closes#1695
Change-Id: I743b890b83903d9e0b3248b032844e7e357d2307
We have long switched to having a single primary global poller, so
there's no point in passing the poller pointer around and storing it in
associated objects. Remove all remnants of non-global pollers.
Change-Id: I5a3fd217d5de51e839e2b04fec7e23643ee83631
This is a dependency of the poller and so is needed in lib/ as well to
make it usable.
Also consolidate the type.
Change-Id: I70ec8a200d6cd65710ac93636a9495cf24c35ef4
All timers have been moved to their dedicated timer threads, making this
mechanism obsolete.
The only victim is the timeout handling for TCP control streams. Since
other TCP streams aren't using timeout handling either, and the TCP
control socket is barely used by anyone, we can live with not having a
dedicated timeout for these streams for now.
Change-Id: I83d9b9a844f4f494ad37b44f5d1312f272beff3f
Move the `ice_slow_timer()` functionality to a separate thread,
so that we do the work more efficiently, and not be dependent
on the call_timer runs by poller.
Furthermore it makes more sense to keep in the `ice.c`,
since it obviously has to do with ice timing.
Additionally:
Update the test-stats due to these changes in the `call_timer()`
We have to call the `ice_slow_timer()` now explicitely from
the test-stats.c, because the `call_timer()` is not anymore
responsible for providing stats counters rate calculations.
Change-Id: I03377dd59ea71c27497e1f4d30164075f05165cd
To do the work more efficiently,
and not be dependent on the call_timer runs by poller,
we should move the ports iterations (stats update from the kernel)
functionality to a separate thread, to make it faster and
not be dependent on what happens in the `call_timer` at all.
Since it has nothing to do with the call timers.
As an additional benefit: we unload the `call_timer` runner.
Change-Id: I511529ce504ef3d29f4e9d6d731ffd470d78d27a
To do the work more efficiently,
and not be dependent on the call_timer runs by poller,
we should move the `stats_rate_min_max()` to a separate thread,
to make it faster and not be dependent on what happens in the `call_timer` at all.
Since it has nothing to do with the call timers.
Change-Id: I9a39e1b63cb8741377f5af5b2d52d4f8b428a0ad
To do the work more efficiently and not be dependent on
the `call_timer` runs by poller, we should move
the releasing of sockets to a separate thread, to make it
faster and not be dependent on what happens in the `call_timer`
at all. Since it has nothing to do with the call timers.
Since now we have two queues:
- thread scope (local): ports_to_release
- global one: ports_to_release_glob
`sockets_releaser()` uses the ports_to_release_glob,
meanwhile appending in the `call_timer()` happens using the
ports_to_release.
Change-Id: Iadd966ac895b2dd64f81269d4fdf5d83747fe0b7
Richard Fuchs
Donat Zenichev
Lucian Balaceanu
Richard Fuchs
Andrii Pogrebennyk
Dennis Yurasov
Orgad Shaneh
Tom Briden