During some performance testing of Asterisk with AGI, ARI, and lots of Local
channels, we noticed that there's quite a hit in performance during channel
creation and releasing to the dialplan (ARI continue). After investigating
the performance spike that occurs during channel creation, we discovered
that we create a lot of channel snapshots that are technically unnecessary.
This includes creating snapshots during:
* AGI execution
* Returning objects for ARI commands
* During some Local channel operations
* During some dialling operations
* During variable setting
* During some bridging operations
And more.
This patch does the following:
- It removes a number of fields from channel snapshots. These fields were
rarely used, were expensive to have on the snapshot, and hurt performance.
This included formats, translation paths, Log Call ID, callgroup, pickup
group, and all channel variables. As a result, AMI Status,
"core show channel", "core show channelvar", and "pjsip show channel" were
modified to either hit the live channel or not show certain pieces of data.
While this is unfortunate, the performance gain from this patch is worth
the loss in behaviour.
- It adds a mechanism to publish a cached snapshot + blob. A large number of
publications were changed to use this, including:
- During Dial begin
- During Variable assignment (if no AMI variables are emitted - if AMI
variables are set, we have to make snapshots when a variable is changed)
- During channel pickup
- When a channel is put on hold/unhold
- When a DTMF digit is begun/ended
- When creating a bridge snapshot
- When an AOC event is raised
- During Local channel optimization/Local bridging
- When endpoint snapshots are generated
- All AGI events
- All ARI responses that return a channel
- Events in the AgentPool, MeetMe, and some in Queue
- Additionally, some extraneous channel snapshots were being made that were
unnecessary. These were removed.
- The result of ast_hashtab_hash_string is now cached in stasis_cache. This
reduces a large number of calls to ast_hashtab_hash_string, which reduced
the amount of time spent in this function in gprof by around 50%.
#ASTERISK-23811 #close
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/3568/
........
Merged revisions 416211 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@416216 65c4cc65-6c06-0410-ace0-fbb531ad65f3
* Remove unused RAII_VAR() declarations. The compiler cannot catch these
because the cleanup function "references" the unused variable. Some
actually allocated and released resources that were never used.
* Fixed some whitespace issues in stasis_bridges.c.
........
Merged revisions 412399 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@412400 65c4cc65-6c06-0410-ace0-fbb531ad65f3
* Fix memory leak in ast_unreal_new_channels(). Made it generate the ;2
uniqueid on a stack variable instead of mallocing it.
* Made send error response to ARI and AMI requests instead of just logging
excessive uniqueid length and allowing truncation. action_originate() and
ari_channels_handle_originate_with_id().
* Fixed minor truncating uniqueid hole when generating the ;2 uniqueid
string length. Created public and internal lengths of uniqueid. The
internal length can handle a max public uniqueid plus an appended ;2.
* free() and ast_free() are NULL tolerant so they don't need a NULL test
before calling.
* Made use better struct initialization format instead of the position
dependent initialization format. Also anything not explicitly initialized
in the struct is initialized to zero by the compiler.
* Made ast_channel_internal_set_fake_ids() use the safer
ast_copy_string() instead of strncpy().
Review: https://reviewboard.asterisk.org/r/3371/
........
Merged revisions 410949 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@410950 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Much needed was a way to assign id to objects on creation, and
much change was necessary to accomplish it. Channel uniqueids
and linkedids are split into separate string and creation time
components without breaking linkedid propgation. This allowed
the uniqueid to be specified by the user interface - and those
values are now carried through to channel creation, adding the
assignedids value to every function in the chain including the
channel drivers. For local channels, the second channel can be
specified or left to default to a ;2 suffix of first. In ARI,
bridge, playback, and snoop objects can also be created with a
specified uniqueid.
Along the way, the args order to allocating channels was fixed
in chan_mgcp and chan_gtalk, and linkedid is no longer lost as
masquerade occurs.
(closes issue ASTERISK-23120)
Review: https://reviewboard.asterisk.org/r/3191/
........
Merged revisions 410157 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@410158 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This patch tweaks the behaviour of POST /channels with channel variables such
that the variables are passed into the pbx.c routines that perform the
origination. This allows the variables to be assigned to the newly created
channels immediately upon their construction, as opposed to be assigned after
the originate has completed.
The upshot of this is that the variables are available on the channels if
they execute in the dialplan, as opposed to only being available once the
channels are answered.
Review: https://reviewboard.asterisk.org/r/3183/
........
Merged revisions 407937 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@407940 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This adds back in support for specifying channel variables during an
originate without compromising the ability to specify query parameters
in the JSON body. This was accomplished by generating the body-parsing
code in a separate function instead of being integrated with the URI
query parameter parsing code such that it could be called by paths with
body parameters. This is transparent to the user of the API and
prevents manual duplication of code or data structures.
(closes issue ASTERISK-23051)
Review: https://reviewboard.asterisk.org/r/3122/
Reported by: Matt Jordan
........
Merged revisions 406003 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@406006 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This change fixes a few memory leaks that were found based
on a mailing list post.
1. Some JSON response messages were never freed. This was
caused by the documentation stating that message references
were stolen when in reality they were not. The code now follows
the documentation and usage has been updated.
2. HTTP response headers were never freed.
3. The variable list for wildcards paths was never freed.
(closes issue ASTERISK-23128)
Reported by: Kenneth Watson (on list)
Review: https://reviewboard.asterisk.org/r/3119/
........
Merged revisions 405325 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@405326 65c4cc65-6c06-0410-ace0-fbb531ad65f3
When we added support for specifying channel variables for an
origination, we didn't consider how that would interact with another
feature, namely specifying request parameters in a JSON request body.
The method of specifying channel variables (as a flat JSON object passed
in the JSON body) interferes with parsing parameters out of the request
body.
Unfortunately, fixing this would be a backward incompatible change. In
the interest of keeping the API sane and keeping our release schedule,
we're dropping the feature for specifying channel variables in the
origination request.
We will bring the feature back soon, as a backward compatible addition
to the API.
(closes issue ASTERISK-23051)
Review: https://reviewboard.asterisk.org/r/3088
........
Merged revisions 404509 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@404512 65c4cc65-6c06-0410-ace0-fbb531ad65f3
When creating channels via ARI, the current code fails to provide any default
format capabilities. For non-virtual channels this isn't really a problem -
the channels typically receive their capabilities as a result of the
underlying channel driver configuration. For virtual channels (such as Local
channels), the lack of any format capabilities causes the Asterisk core to
make some 'odd' choices with respect to the translation paths. The issue
reporter had some paths that had 3 hops on each channel leg, causing multiple
transcodings and some really crappy audio/performance.
By specifying a baseline of SLIN, we prevent that from occurring. Note that
this is what AMI does when it performs an Originate, as does res_clioriginate.
Review: https://reviewboard.asterisk.org/r/3068/
(issue ASTERISK-22962)
Reported by: Matt DiMeo
........
Merged revisions 403993 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@403994 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Added the ability to specify channel variables when creating/originating a
channel in ARI. The variables are sent in the body of the request and should
be formatted as a single level JSON object. No nested objects allowed.
For example: {"variable1": "foo", "variable2": "bar"}.
(closes issue ASTERISK-22872)
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/3052/
........
Merged revisions 403752 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@403757 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This change prevents channels used as implementation details from
leaking out to ARI. It does this by preventing creation of JSON blobs
of channel snapshots created from those channels and sanitizing JSON
blobs of bridge snapshots as they are created. This introduces a
framework for excluding information from output targeted at Stasis
applications on a consumer-by-consumer basis using channel sanitization
callbacks which could be extended to bridges or endpoints if necessary.
This prevents unhelpful error messages from being generated by
ast_json_pack.
This also corrects a bug where BridgeCreated events would not be
created.
(closes issue ASTERISK-22744)
Review: https://reviewboard.asterisk.org/r/2987/
Reported by: David M. Lee
........
Merged revisions 403069 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@403070 65c4cc65-6c06-0410-ace0-fbb531ad65f3
While working on building client libraries from the Swagger API, I
noticed a problem with the nicknames.
channel.deleteChannel()
channel.answerChannel()
channel.muteChannel()
Etc. We put the object name in the nickname (since we were generating C
code), but it makes OO generators redundant.
This patch makes the nicknames more OO friendly. This resulted in a lot
of name changing within the res_ari_*.so modules, but not much else.
There were a couple of other fixed I made in the process.
* When reversible operations (POST /hold, POST /unhold) were made more
RESTful (POST /hold, DELETE /unhold), the path for the second operation
was left in the API declaration. This worked, but really the two
operations should have been on the same API.
* The POST /unmute operation had still not been REST-ified.
Review: https://reviewboard.asterisk.org/r/2940/
........
Merged revisions 402528 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@402529 65c4cc65-6c06-0410-ace0-fbb531ad65f3
The ring operation sends ringing to the specified channel it is invoked on.
The dtmf operation can be used to send DTMF digits to the specified channel
of a specific length with a wait time in between. Finally hangup reasons
allow you to specify why a channel is being hung up (busy, congestion).
Early media behavior has also been tweaked slightly. When playing media to a channel
it will no longer automatically answer. If it has not been answered a progress indication
is sent instead.
(closes issue ASTERISK-22701)
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/2916/
........
Merged revisions 402358 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@402359 65c4cc65-6c06-0410-ace0-fbb531ad65f3
If a Stasis application is specified an implicit subscription is done on the originated
channel. This was previously done with the channel lock held which is dangerous as the
underlying code locks the container and iterates items. This change releases the lock
on the originated channel before subscribing occurs.
(closes issue ASTERISK-22768)
Reported by: Matt Jordan
Review: https://reviewboard.asterisk.org/r/2979/
........
Merged revisions 402346 from http://svn.asterisk.org/svn/asterisk/branches/12
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@402347 65c4cc65-6c06-0410-ace0-fbb531ad65f3
Both /asterisk/variable and /channel/{channelId}/variable requires a
?variable parameter to be passed into the query. But we weren't checking
for the parameter being missing, which caused a segfault.
All calls now properly return 400 Bad Request errors when the parameter
is missing. The Swagger api-docs were updated accordingly.
(closes issue ASTERISK-22273)
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@397306 65c4cc65-6c06-0410-ace0-fbb531ad65f3
In working with res_stasis, I discovered a significant limitation to
the current structure of stasis_caching_topics: you cannot subscribe
to cache updates for a single channel/bridge/endpoint/etc.
To address this, this patch splits the cache away from the
stasis_caching_topic, making it a first class object. The stasis_cache
object is shared amongst individual stasis_caching_topics that are
created per channel/endpoint/etc. These are still forwarded to global
whatever_all_cached topics, so their use from most of the code does
not change.
In making these changes, I noticed that we frequently used a similar
pattern for bridges, endpoints and channels:
single_topic ----------------> all_topic
^
|
single_topic_cached ----+----> all_topic_cached
|
+----> cache
This pattern was extracted as the 'Stasis Caching Pattern', defined in
stasis_caching_pattern.h. This avoids a lot of duplicate code between
the different domain objects.
Since the cache is now disassociated from its upstream caching topics,
this also necessitated a change to how the 'guaranteed' flag worked
for retrieving from a cache. The code for handling the caching
guarantee was extracted into a 'stasis_topic_wait' function, which
works for any stasis_topic.
(closes issue ASTERISK-22002)
Review: https://reviewboard.asterisk.org/r/2672/
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@395954 65c4cc65-6c06-0410-ace0-fbb531ad65f3
This renames all files and API calls from several variants of
Stasis-HTTP to ARI including:
* Stasis-HTTP -> ARI
* STASIS_HTTP -> ARI
* stasis_http -> ari (ast_ari for global symbols, file names as well)
* stasis http -> ARI
Review: https://reviewboard.asterisk.org/r/2706/
(closes issue ASTERISK-22136)
git-svn-id: https://origsvn.digium.com/svn/asterisk/trunk@395603 65c4cc65-6c06-0410-ace0-fbb531ad65f3
