The following scenario disclouse an issue with "ngcp_clean" lua
function in case of callee peer:
- A is a local subscriber with Peer_auth set and
force_outbound_call_to_peer set;
- B is a local subscriber with Peer_auth set and
force_outbound_call_to_peer set;
A calls B.
We clean and load callee preferences of B.
Call is detected as forced to PSTN.
We clean and load peer preferences.
When we set outbound peer_auth* values, the $xavp(callee_prefs)
still contains values from B (instead of peering preferences).
When we clean preferences for the callee peer, we only
clean "callee_peer_prefs" and not "callee_prefs".
"callee_prefs" are clean only in case of 'usr' or 'dom'
preferences.
This patch try to fix this problem in order to fix this
scenario with outbound user authentication.
Change-Id: I76dfa6f9e074b3d83ac67bd818e59233ca50492c
(cherry picked from commit 8cdd63731b)
* we are checking that the variable is not nil but on
start we were setting the value to an empty metatable
Change-Id: Ib5b409b5c6e837baef38c35c7e619a46f44d51b9
(cherry picked from commit 3a008159b2)
(cherry picked from commit a1a7bc3da4)
The 'duration' variable is a string but it was written as int,
causing kamailio to write error lines
Change-Id: I2df1524d331e9b1d54bb0bdb84c06f4280e14fae
(cherry picked from commit d81fe7778d)
> ngcp/fp.lua:30:121: line is too long (126 > 120)
> ngcp/pprof.lua:28:121: line is too long (158 > 120)
Change-Id: Icf87f29049b2cc4ce29e7c1f5c78d9f2a53572c0
(cherry picked from commit 41dd39627e)
https://www.lua.org/manual/5.2/manual.html#8.3
> (Modules are not expected to set global variables anymore.)
EXPORT_ASSERT_TO_GLOBALS was just there for backwards compatibility
Change-Id: I11ac6d24746ca3c965b16ad523efff1f12ef0e9a
We are going to use both modules app_lua and app_lua_sr until we
solve the missing pieces on KEMI:
* KSR.pvx.xavp_get() returns a string like '<<xavp:%p>>' not a table like
sr.xavp.get()
* KEMI lacks of something like sr.xavp.get_keys() funtion
So, migrate everything that we can to KEMI and keep using sr approach
for just that two cases
- mocks/sr.lua: use same objects for same behaviour
Change-Id: I475ff8f820586cdc94c75bff7466238215e05673
The script collects all the callID from redis, then, one by one
it checks if callID exist in kamailo. At this time a call existing
in redis before may be destroyed in kamailio dlg.list. And so
script believe it's a stuck call.
Solution: collect CallID from REDIS and kamailio.dlg at the same
time, and compare afterwards.
Change-Id: I233a498e42bf5113b5b11ea688487e1840cd84a6
We also need to switch from python-xmlrunner to python3-pytest, as
the former is not present in Debian unstable/testing anymore (see
<https://bugs.debian.org/935688>), and does not provide python3
support in Debian buster anyway.
Change-Id: I8af9ead2755af2e24f5bedf53d8cd20c4ed63b85
Merge two file stanzas for Sipwise copyright, and use the one with the
canonical Sipwise copyright entity.
Change-Id: Iae16419b8d7d6af0555968e109a777b7951c121c
In the new version of LuaUnit assertTrue() and assertFalse()
functions are changed to striclty evaluate that the data is
True or false.
Where necessary the fuctions have been substituted in the test
with the new assertEvalToTrue() or assertNil().
Change-Id: I0c728323752a4458e93f516c7529670433c516d1
Sipwise Jenkins Builder
Daniel Grotti
Victor Seva
Marco Capetta
Guillem Jover
Alexander Lutay