ben.carrasco/freeswitch
1
0
Fork 0
mirror of https://github.com/signalwire/freeswitch.git synced 2026-07-24 13:12:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

FS-9952: Committing to show problem with ks_pool_resize

Browse Source
This commit is contained in:
Shane Bryldt
2017-02-15 00:56:29 +00:00
committed by Mike Jerris
parent d6d8ede6b9
commit a7add33519
7 changed files with 269 additions and 136 deletions
Download Patch File Download Diff File Expand all files Collapse all files
libs/libblade/src/blade_stack.c
+34 -7
View File
@@ -44,11 +44,12 @@ struct blade_handle_s {
ks_pool_t *pool;
ks_thread_pool_t *tpool;
config_setting_t *config_service;
config_setting_t *config_directory;
config_setting_t *config_datastore;
ks_hash_t *transports;
blade_identity_t *identity;
blade_datastore_t *datastore;
};
@@ -164,7 +165,7 @@ KS_DECLARE(ks_status_t) blade_handle_destroy(blade_handle_t **bhP)
ks_status_t blade_handle_config(blade_handle_t *bh, config_setting_t *config)
{
config_setting_t *service = NULL;
config_setting_t *directory = NULL;
config_setting_t *datastore = NULL;
ks_assert(bh);
@@ -172,13 +173,13 @@ ks_status_t blade_handle_config(blade_handle_t *bh, config_setting_t *config)
if (!config) return KS_STATUS_FAIL;
if (!config_setting_is_group(config)) return KS_STATUS_FAIL;
service = config_setting_get_member(config, "service");
directory = config_setting_get_member(config, "directory");
datastore = config_setting_get_member(config, "datastore");
//if (datastore && !config_setting_is_group(datastore)) return KS_STATUS_FAIL;
bh->config_service = service;
bh->config_directory = directory;
bh->config_datastore = datastore;
return KS_STATUS_SUCCESS;
@@ -201,7 +202,11 @@ KS_DECLARE(ks_status_t) blade_handle_startup(blade_handle_t *bh, config_setting_
return KS_STATUS_FAIL;
}
}
// @todo load DSOs
// @todo call onload and onstartup callbacks for modules from DSOs
return KS_STATUS_SUCCESS;
}
@@ -209,9 +214,12 @@ KS_DECLARE(ks_status_t) blade_handle_shutdown(blade_handle_t *bh)
{
ks_assert(bh);
// @todo cleanup registered transports
// @todo call onshutdown and onunload callbacks for modules from DSOs
// @todo unload DSOs
if (blade_handle_datastore_available(bh)) blade_datastore_destroy(&bh->datastore);
return KS_STATUS_SUCCESS;
}
@@ -279,14 +287,33 @@ KS_DECLARE(ks_status_t) blade_handle_connect(blade_handle_t *bh, blade_connectio
ks_assert(bh);
ks_assert(target);
// @todo this should take a callback, and push this to a queue to be processed async from another thread on the handle
// which will allow the onconnect callback to block while doing things like DNS lookups without having unknown
// impact depending on the caller thread
ks_hash_read_lock(bh->transports);
blade_identity_parameter_get(target, "transport", &tname);
tname = blade_identity_parameter_get(target, "transport");
if (tname) {
bhtr = ks_hash_search(bh->transports, (void *)tname, KS_UNLOCKED);
if (!bhtr) {
// @todo error logging, target has an explicit transport that is not available in the local transports registry
// discuss later whether this scenario should still attempt other transports when target is explicit
// @note discussions indicate that by default messages should favor relaying through a master service, unless
// an existing direct connection already exists to the target (which if the target is the master node, then there is
// no conflict of proper routing). This also applies to routing for identities which relate to groups, relaying should
// most often occur through a master service, however there may be scenarios that exist where an existing session
// exists dedicated to faster delivery for a group (IE, through an ampq cluster directly, such as master services
// syncing with each other through a pub/sub). There is also the potential that instead of a separate session, the
// current session with a master service may be able to have another connection attached which represents access through
// amqp, which in turn acts as a preferred router for only group identities
// This information does not directly apply to connecting, but should be noted for the next level up where you simply
// send a message which will not actually connect, only check for existing sessions for the target and master service
// @note relaying by master services should take a slightly different path, when they receive something not for the
// master service itself, it should relay this on to all other master services, which in turn all including original
// receiver pass on to any sessions matching an identity that is part of the group, alternatively they can use a pub/sub
// like amqp to relay between the master services more efficiently than using the websocket to send every master service
// session the message individually
}
} else {
for (ks_hash_iterator_t *it = ks_hash_first(bh->transports, KS_UNLOCKED); it; it = ks_hash_next(&it)) {