====== Chat log IRC Devel Meeting - 2014-05-14 ====== ===== outstanding issues at this time, if any ===== [2014-05-14 16:06:41] apparently is a race for cleaning up dialog which goes from initial state to terminated state\\ [2014-05-14 16:06:44] reported recentrly\\ [2014-05-14 16:07:04] i looked a bit, couldn’t reproduce, but I expect is under high load\\ [2014-05-14 16:07:25] as a lead now, could be due to pre/post-script callbacks\\ [2014-05-14 16:07:25] I also brought up the topic of CONFIRMED_NA dialogs getting stuck until dialog_timeout, but with no response (that I saw).\\ [2014-05-14 16:08:06] abalashov: yes, together with delayed release of event route reported by osas\\ [2014-05-14 16:08:16] will make a bit of work on dialog\\ [2014-05-14 16:08:41] which might end up in getting rid of dependency on tm … which is not that useful, imo\\ [2014-05-14 16:08:59] dialog can handle the requests and replies by itself\\ [2014-05-14 16:09:04] i will see\\ [2014-05-14 16:09:15] Agree that knowledge of transaction state is not really essential to dialog functionality; it's just convenient.\\ [2014-05-14 16:09:35] it is not exported to dialog api in any way\\ [2014-05-14 16:09:45] Doesn't other modules that depend on dialog depend on transaction state?\\ [2014-05-14 16:09:48] Like dialog\\ [2014-05-14 16:09:48] just convenience in matching replies some time\\ [2014-05-14 16:09:50] And can sometimes break things -- if dialog-altering sequential request comes through but in some way improperly (say, no Route set) and TM doesn't trigger the right callbacks / doesn't know about the transaction.\\ [2014-05-14 16:10:15] oej: what do you mean with ‘Like dialog’?\\ [2014-05-14 16:10:22] http://kamailio.org/docs/modules/devel/modules/pua_dialoginfo.html \\ [2014-05-14 16:10:25] Depends on dialog\\ [2014-05-14 16:10:35] ahh, ok, pua_dialoginfo\\ [2014-05-14 16:10:43] And dialoginfo shows transaction states of the invite trans\\ [2014-05-14 16:10:46] no, they don’t depend on transaction state\\ [2014-05-14 16:10:56] only on dialog states\\ [2014-05-14 16:10:59] according to the spec it does\\ [2014-05-14 16:11:18] well, the states are initial, ringing, confirmed\\ [2014-05-14 16:11:46] dialog relies on tm for getting back replies\\ [2014-05-14 16:11:46] which correspond to some of transaction states, but in the code, dialog doesn’t propagate tm info\\ [2014-05-14 16:12:57] no, it doesn't\\ [2014-05-14 16:13:06] miconda, it seems I neglected to see your reply on the CONFIRMED_NA topic ... http://lists.sip-router.org/pipermail/sr-dev/2014-April/023862.html -- sorry for the misapprehension. \\ [2014-05-14 16:13:58] miconda: The proxy does get BYEs in this scenario, but because the dialog is not yet confirmed+established they do not result in tearing down the dialog.\\ [2014-05-14 16:14:11] abalashov: ok\\ [2014-05-14 16:14:36] miconda: This generally happens due to 200 OK-CANCEL race, but can also happen due to malformed ACKs and whatnot.\\ [2014-05-14 16:14:41] found another one recently, when the dialog was not destroyed on timeout if bye failed to be sent\\ [2014-05-14 16:15:05] to conclude, dialog needs a bit of hammering :-)\\ [2014-05-14 16:15:07] Dialog is a very messy piece of the proxy, it just happens to be very extremely useful in the applied dimension.\\ [2014-05-14 16:16:30] other issues?\\ [2014-05-14 16:17:30] miconda: I hear you on setting the dialog timeout to one timer prior to ACK and another after, but maybe there should be a separate timeout setting for unconfirmed dialogs? Or is that asking too much?\\ [2014-05-14 16:17:40] miconda: It's an extremely common problem in high-pace environments due to 200 OK/CANCEL race.\\ [2014-05-14 16:18:17] abalashov: separate timeout can be added indeed\\ [2014-05-14 16:18:26] miconda: I'd be willing to work on that.\\ [2014-05-14 16:18:36] miconda: Problem is it's so difficult to simulate a 200 OK/CANCEL race deliberately.\\ [2014-05-14 16:19:07] miconda: I might take the approach of sending some bullshit ACK instead. Maybe with the wrong From tag. That reliably works, I know from unfortunate personal experience with the VOS3000.\\ [2014-05-14 16:19:20] abalashov: we will give you a free license for devel purposes\\ [2014-05-14 16:19:21] (or rather, wrong To tag, actually)\\ [2014-05-14 16:19:22] :-)\\ [2014-05-14 16:20:13] sorry for asking but what is the difference between dialog and dialog_ng?\\ [2014-05-14 16:20:26] That's a good question. I am unclear on this as well.\\ [2014-05-14 16:20:51] dialog_ng - dialog = _ng\\ [2014-05-14 16:20:59] Hah!\\ [2014-05-14 16:21:09] _ng version is an attempt to rework dialog\\ [2014-05-14 16:21:11] linuxmaniac: In theory it is supposed to be a next generation dialog module that deals better with spirals and forking.\\ [2014-05-14 16:21:19] linuxmaniac: In practice I am unsure as to what the practical difference is at this point.\\ [2014-05-14 16:21:21] it was a discussion some time ago\\ [2014-05-14 16:21:41] but apparently dialog_ng doesn’t have all features of dialog\\ [2014-05-14 16:21:50] Right, so in practice I don't see it being so usable.\\ [2014-05-14 16:21:59] there is a wiki page about the discusion on dialog_ng\\ [2014-05-14 16:22:08] but, is anyone working on it?\\ [2014-05-14 16:22:19] We are making heavy use of dialog_ng for IMS topics.\\ [2014-05-14 16:22:22] and, if i got it right, is mostly ims oriented now\\ [2014-05-14 16:22:26] the ims devs\\ [2014-05-14 16:22:32] Here we are!\\ [2014-05-14 16:22:55] any plans to get all the features of dialog?\\ [2014-05-14 16:23:01] Nevertheless, this is likely to be confusing to the general public and newbies and so on.\\ [2014-05-14 16:23:01] by Jason Penton: "Dialog2 is a start at the following: http://www.kamailio.org/dokuwiki/doku.php/modules-new-design:dialog-module-design We needed this functionality for the IMS modules. In IMS a proxy can take the same session in both terminating and originating modes (ie 2 dialogs for the same session). dialog 1 does not support this. Dialog2 has been tested to at least support the functionality we required and we will build \\ [2014-05-14 16:23:03] so carstenbock can say the diff between dialog and dialog_ng\\ [2014-05-14 16:23:13] dialog2 was the original name of dialog_ng\\ [2014-05-14 16:23:45] My personal reading when I first ran into it was, "Oh, it's _ng, that sounds like some avant garde stuff, I don't think I'll use it." But in the case of rtpproxy_ng this would be incorrect thinking. :-)\\ [2014-05-14 16:23:59] :-)\\ [2014-05-14 16:24:17] avant garde\\ [2014-05-14 16:24:18] now we don't have any _ng\\ [2014-05-14 16:24:25] Yeah, I know.\\ [2014-05-14 16:24:34] rtpengine_ng - specially for IMS :-)\\ [2014-05-14 16:24:45] At the moment the development on the dialog_ng module is on "per demand" basis.\\ [2014-05-14 16:24:58] AFAIR, the only missing thing in dialog_ng, is the full database support\\ [2014-05-14 16:25:15] so there is no intension to replace dialog with dialog_ng in the long term\\ [2014-05-14 16:25:30] ... contrary to the triumphant announcement in the dialog_ng documentation ...\\ [2014-05-14 16:26:26] I'm just asking because it seems dialog need some work\\ [2014-05-14 16:26:30] maybe in a future. Currently I believe it would be impractical to do that.\\ [2014-05-14 16:26:58] OK, moving on\\ [2014-05-14 16:27:33] caruizdiaz: I think the dialog_ng module should perhaps be amended to take a more moderate stance, then. As it is written now, it states, "El nuevo dialog! Viva la revolucion!"\\ [2014-05-14 16:27:42] caruizdiaz: + documentation\\ [2014-05-14 16:28:03] +7\\ [2014-05-14 16:28:48] Maybe an option would be, to rename dialog_ng to dialog_ims, to reduce the confusion.\\ [2014-05-14 16:29:09] (at least as long as it's mostly used for IMS)\\ [2014-05-14 16:29:09] abalashov: jajajaj, maybe yes, but I don't think it is affecting the new comers. I did not read any inquiry or doubt about which of them to use.\\ [2014-05-14 16:29:26] True...\\ [2014-05-14 16:29:31] ok, so are we done?\\ [2014-05-14 16:29:34] We had it in a release so it's late for rename\\ [2014-05-14 16:29:43] with dlg vs dlg_ng …\\ [2014-05-14 16:29:46] But improving the docs is a good thing to avoid any confusion\\ [2014-05-14 16:29:52] Okay, agreed.\\ [2014-05-14 16:30:11] done\\ [2014-05-14 16:30:12] Assigned to carstenbock aka "The IMS Guy"\\ [2014-05-14 16:30:23] Damn!\\ [2014-05-14 16:30:24] :-)\\ [2014-05-14 16:30:26] :D:D\\ [2014-05-14 16:30:42] :)\\ ===== major release ===== [2014-05-14 16:31:04] most probably autumn time frame\\ [2014-05-14 16:31:55] it fits the development cycle interval and I guess summer is not the ideal for testing a frozen branch\\ [2014-05-14 16:32:26] perhaps we can decide the date to freeze development at the end of summer, via mailing list\\ [2014-05-14 16:32:39] +1\\ [2014-05-14 16:32:44] What's new?\\ [2014-05-14 16:32:51] Event stuff\\ [2014-05-14 16:32:52] i think there are like 3 new modules now\\ [2014-05-14 16:32:53] Will it be Kamailio 5.0? :) Or is it time for it to become Kamailio Vista?\\ [2014-05-14 16:33:11] plus quite a lot in other old modules\\ [2014-05-14 16:33:12] :D\\ [2014-05-14 16:33:20] Kamailio 95\\ [2014-05-14 16:33:27] Inspired by SAAB 95\\ [2014-05-14 16:33:29] afaik, there are 3-4 modules to be added soon …\\ [2014-05-14 16:33:30] (of course)\\ [2014-05-14 16:33:33] What's that?\\ [2014-05-14 16:33:36] Kamailio_NG\\ [2014-05-14 16:34:00] two erlang modules are on github in a different repo\\ [2014-05-14 16:34:08] Actually, I assume it will be Kamailio 4.2. Right?\\ [2014-05-14 16:34:19] i discussed with the developer, he wants more polishing before incorporating them\\ [2014-05-14 16:34:32] Erlang? Cool.\\ [2014-05-14 16:34:45] do you have the links?\\ [2014-05-14 16:35:04] one module by kazoo guys to be reviewed and adjusted to be more generic if it is the case (on github as well)\\ [2014-05-14 16:35:25] caruizdiaz: searching on github should reveal them\\ [2014-05-14 16:35:28] No Go yet?\\ [2014-05-14 16:35:38] ok, tks\\ [2014-05-14 16:35:56] linuxmaniac: No indeed, and neither Haskell.\\ [2014-05-14 16:36:12] the developer of json modules said at some point in time that they refactored, but i haven’t seen any updates yet\\ [2014-05-14 16:36:40] miconda: Has there been any movement on the topic of how exactly the evapi should work?\\ [2014-05-14 16:36:57] abalashov: “at your will”\\ [2014-05-14 16:37:03] no format enforced now\\ [2014-05-14 16:37:12] What about subscriptions and filtering?\\ [2014-05-14 16:38:25] nothing yet — looks like a busy summer for you :-)\\ [2014-05-14 16:38:34] :-)\\ [2014-05-14 16:38:53] Previously I had done all this with uac_req_send().\\ [2014-05-14 16:38:57] This offers a novel approach.\\ [2014-05-14 16:39:29] to conclude, I’m sure there will be plenty of new modules by the time of the next major release\\ [2014-05-14 16:39:38] Cool\\ [2014-05-14 16:40:01] those are what I am aware at this time to be on pipeline\\ [2014-05-14 16:40:16] Are there are any defunct and unmaintained modules that should be deprecated?\\ [2014-05-14 16:40:39] I have my eye on 'seas', not sure what it's status is.\\ [2014-05-14 16:40:41] and maybe 1-2 more from me … let’s see how the time goes\\ [2014-05-14 16:40:44] *its\\ [2014-05-14 16:40:54] seas should work ok\\ [2014-05-14 16:41:07] hi\\ [2014-05-14 16:41:08] iirc, some stuff related to tcp was missing there\\ [2014-05-14 16:41:19] but for udp, should be no problems\\ [2014-05-14 16:41:58] I get lots of questions about fleshing out the Kamailio SNMP MIB with something more useful, since bellheads love SNMP.\\ [2014-05-14 16:42:08] and, being asked up here, it’s going to be 4.2.0\\ [2014-05-14 16:42:13] I always just tell them to wrap custom OIDs with kamctl queries.\\ [2014-05-14 16:42:38] 5.0 when we deprecate ipv4 and we support ipv6 and ipv9\\ [2014-05-14 16:42:48] aka Kamailio Vista\\ [2014-05-14 16:43:07] erlang module is at github.com/stangrze/sip-router-erlang-module\\ [2014-05-14 16:43:45] abalashov: should not be hard to add to snmpstats, oej did latel several of them\\ [2014-05-14 16:43:59] stangrze: hello! thanks for the link to erlang module\\ [2014-05-14 16:44:00] its quite usable now, would appreciate is someone dares to test it wit dome complex config\\ [2014-05-14 16:44:01] topic suggestion: modifying/enhancing the db api to support nosql databases\\ [2014-05-14 16:44:22] stangrze: i was just going to ask abotu the state\\ [2014-05-14 16:44:25] ...some..\\ [2014-05-14 16:44:49] stangrze: going to see if I can hunt some erlang fans\\ [2014-05-14 16:45:21] stangrze: perhaps testing will come easier for everyone when the module gets into main repo\\ [2014-05-14 16:45:42] database interface for srdb1 is quite complete, you can use boss_db app to connect to mysql database, and it should doo all stuff\\ [2014-05-14 16:46:26] ok, let’s conclude with major release: autumn, decision on date by end of summer\\ [2014-05-14 16:47:14] i got busy with broadcom's as-feature-event stuff i need for my work, than i'll get back to erlang\\ [2014-05-14 16:47:39] stangrze: You mean Broadsoft? :)\\ [2014-05-14 16:47:53] broadsoft, sory ;-)\\ ===== minor release ===== [2014-05-14 16:49:00] planning one by end of may, early june\\ [2014-05-14 16:49:13] before starting the holidays season\\ [2014-05-14 16:50:00] most probably for both latest stable branches this time — 4.0 and 4.1\\ [2014-05-14 16:50:18] so this was short — I gues …\\ [2014-05-14 16:50:27] we get back to proposed topics later\\ [2014-05-14 16:50:36] jessie will freeze at 6th of November 2014\\ [2014-05-14 16:50:51] just to finish what we have on wiki\\ [2014-05-14 16:51:07] linuxmaniac: so we have to do a major release before to get it in, right?\\ [2014-05-14 16:51:14] yes\\ [2014-05-14 16:51:16] I will freeze in october already...\\ [2014-05-14 16:51:19] ok\\ [2014-05-14 16:51:24] :-)\\ [2014-05-14 16:51:36] hello\\ [2014-05-14 16:51:41] and afer that long maintain work\\ [2014-05-14 16:51:45] xD\\ [2014-05-14 16:52:01] next topic: old or forgotten patches\\ [2014-05-14 16:52:05] just security patches for stable release\\ [2014-05-14 16:52:28] multipart bodies!!! :-(\\ [2014-05-14 16:52:32] jaja\\ [2014-05-14 16:52:43] My patch that deprecates IPv4 and non-TLS connections was forgotten. But it's probably best left forgotten.\\ [2014-05-14 16:53:02] linuxmaniac: should be done soon\\ [2014-05-14 16:53:15] the one for textops is more scarry\\ [2014-05-14 16:53:34] but will be finished soon\\ [2014-05-14 16:53:34] yes, and now I'm working on dialplan \\ [2014-05-14 16:53:57] this will be even scarier \\ [2014-05-14 16:53:57] the one for websockets, Peter responded on mailing list, as he cannot attend here\\ [2014-05-14 16:54:48] my plan for the one on websockets (to get rid of libunistring) is to have defines to select one or another …\\ [2014-05-14 16:55:07] in some OSes is hard to get libunistring\\ [2014-05-14 16:55:20] eZz: hello!\\ ===== mirrors ===== [2014-05-14 16:55:57] +1 to get rid of libunistring\\ [2014-05-14 16:56:33] I have to say that the Makefiles can be improved, I had recently a report because kamailio couldn't build without pkg-config\\ [2014-05-14 16:56:33] we may get some US-located server\\ [2014-05-14 16:56:35] What's the alternative to libunistring\\ [2014-05-14 16:56:51] (sorry, missed "next topic" . MOve on!)\\ [2014-05-14 16:56:54] oej: embedded version\\ [2014-05-14 16:57:04] oej: no problem …\\ [2014-05-14 16:57:32] oej: there was a patch to replace the function used from libunistring with one to be inside the module\\ [2014-05-14 16:57:39] ok\\ [2014-05-14 16:57:48] because it is about conversion from utf8 to ascii, or so ...\\ [2014-05-14 16:57:56] Unicode is not easy, so we need to make sure it works properly\\ [2014-05-14 16:58:18] don’t remember by heart now exactly\\ [2014-05-14 16:58:37] but yes, we will keep both options by compile time to be able to switch in case of problems\\ [2014-05-14 16:58:58] Ok, back to "Mirrors"\\ [2014-05-14 16:59:05] anything else for old patches?\\ [2014-05-14 16:59:15] ok then, back to mirrors\\ [2014-05-14 16:59:23] as said, we may get some US-located server\\ [2014-05-14 16:59:36] thanks to qxork\\ [2014-05-14 16:59:45] Thanks Qxork\\ [2014-05-14 16:59:51] Thanks qxork !\\ [2014-05-14 17:00:00] and I thought of mirroring the debian apt repos\\ [2014-05-14 17:00:01] MIrror of what btw?\\ [2014-05-14 17:00:25] please tell me when will be a free talk, I did not declared any topics\\ [2014-05-14 17:00:27] and build a rpm repo\\ [2014-05-14 17:00:55] oej: so we get a server (virtual)\\ [2014-05-14 17:01:00] and we can use it\\ [2014-05-14 17:01:23] we are ok with apt repos\\ [2014-05-14 17:01:31] but rpms are on open suse build servers\\ [2014-05-14 17:01:49] and doesn’t appear to be easy to get via some package manager\\ [2014-05-14 17:01:55] mika: there is an easy way to build rpm too in jenkins?\\ [2014-05-14 17:02:17] linuxmaniac: nope, i've a buddy working on it, but it needs work\\ [2014-05-14 17:02:25] Ok\\ [2014-05-14 17:02:32] not being a rpm-base distro user, I just thought of cloning the rpms and building a repository out of them\\ [2014-05-14 17:03:10] clean rpm builds similar to what we do with debs would be nice though :) what distris do you have in mind?\\ [2014-05-14 17:03:15] then eventually set a mirror for apt repos as well, although I understood from sipwise guys that the current one is on amazon, so it is distributed already\\ [2014-05-14 17:03:50] miconda: been in amazon doesn't mean distributed\\ [2014-05-14 17:03:50] mika: now we (I) do for centos, redhat, opensuse and fedora\\ [2014-05-14 17:04:03] various versions\\ [2014-05-14 17:04:14] supported by opensuse build service\\ [2014-05-14 17:04:28] linuxmaniac: ok, right, wanted to say more reliable :-)\\ [2014-05-14 17:04:30] miconda: alright\\ [2014-05-14 17:04:59] Gotta leave the office, or they will lock me in. Have a good meeting!\\ [2014-05-14 17:05:20] glad that rpm based distros are taken into account :)\\ [2014-05-14 17:05:38] bye oej :)\\ [2014-05-14 17:05:40] have a great evening, oej!\\ [2014-05-14 17:06:04] for reference: https://build.opensuse.org/project/show/home:kamailio:telephony \\ [2014-05-14 17:06:49] so my question would have been: anyone here knowing to set a yum/whatsoever repository?\\ [2014-05-14 17:06:52] thanks\\ [2014-05-14 17:07:24] caruizdiaz: so you wanted to say that you can take care of?!?\\ [2014-05-14 17:07:56] jajaja, I could, but it will take me some time to get used to the rpm build system.\\ [2014-05-14 17:08:02] can you wait? ;)\\ [2014-05-14 17:08:19] some years ago I was a maintainer of Slackware packages mirror and did it automaticaly\\ [2014-05-14 17:08:28] build service is done, that’s easy\\ [2014-05-14 17:08:56] IIRC createrepo should even work on Debian to provide a yum mirror\\ [2014-05-14 17:08:59] then it would be repo configuration only\\ [2014-05-14 17:09:00] needed: a script to download the rpms and build a yum/etc repo\\ [2014-05-14 17:09:42] ok … let’s get back on this when we have the server\\ [2014-05-14 17:09:54] will discuss on mailing list\\ [2014-05-14 17:10:01] btw, checkinstall can create rpm/debian/slack/etc else packages\\ [2014-05-14 17:10:31] I think peter had done some work on this topic for his centos servers. I will catch up with his work and see how I can help\\ [2014-05-14 17:10:43] ok\\ ===== world wide events ===== [2014-05-14 17:10:53] next topic: world wide events …\\ [2014-05-14 17:11:36] if anyone here is participating to an event and approaches something touching kamailio, we can put news, etc… on site\\ [2014-05-14 17:11:46] I'll add to the docs in the Wiki, how to use the SUSE services as a yum-repo...\\ [2014-05-14 17:12:17] I'm working on 2 new modules according to security\\ [2014-05-14 17:12:23] or if you know an event that should be attended, send and email and we will see if anyone can attend\\ [2014-05-14 17:12:43] eZz: great, wait a bit and we will get on that topic\\ [2014-05-14 17:12:55] miconda: ok\\ [2014-05-14 17:13:40] to conclude on events — Cluecon and Astricon will be on our radar\\ [2014-05-14 17:14:04] what about TADHack?\\ [2014-05-14 17:14:09] but also smaller events are good, those kind of meetups\\ [2014-05-14 17:14:24] caruizdiaz: looks interesting, overlaps with vacation\\ [2014-05-14 17:14:38] but would be good if anyone can participate on site or remote\\ [2014-05-14 17:14:43] it's not vacation in here :D\\ [2014-05-14 17:15:05] here we done one every 10 years, so …\\ [2014-05-14 17:15:05] I think I will. I will let you know if so.\\ [2014-05-14 17:15:06] :-)\\ [2014-05-14 17:15:11] ok\\ [2014-05-14 17:15:17] have to run!\\ [2014-05-14 17:15:32] next topic: ipv6, security\\ [2014-05-14 17:15:43] oej is gone, so it should be short :-)\\ [2014-05-14 17:15:53] xD\\ [2014-05-14 17:16:02] any experiences with IPv6, any issues?\\ [2014-05-14 17:16:16] not much production from my interactions …\\ [2014-05-14 17:17:06] anyhow, for new modules dealing with ip addresses, ipv6 should be considered from early phase\\ [2014-05-14 17:17:24] some providers in germany started to shift to ipv6\\ [2014-05-14 17:17:50] miconda: we have some clients with ipv6\\ [2014-05-14 17:18:03] to conclude: I am not aware of any issue, just wanted to get a round of others’ opinions\\ [2014-05-14 17:18:23] linuxmaniac: ok, so it’s working …\\ [2014-05-14 17:18:37] I not aware of any issue \\ [2014-05-14 17:18:41] do you have any figures (estimations) of subscribers on ipv6\\ [2014-05-14 17:18:49] like order of magnitude\\ [2014-05-14 17:18:57] not me, but maybe Daniel or Andrew\\ [2014-05-14 17:18:59] 1000+, 10 000+ …\\ [2014-05-14 17:19:05] ok\\ ===== security ===== [2014-05-14 17:19:34] We wanted to put IPv6 in a customer network, but then we noticed that basically all fibre-modems and ISDN boxes in that network did not support IPv6 and it's not on the roadmap of the manufacturers :-(\\ [2014-05-14 17:19:57] there is the issue of tls-only listen …\\ [2014-05-14 17:20:06] that should be fixed soon\\ [2014-05-14 17:20:52] before the next major release I will review to see if anything else needs to be addresses in tls module\\ [2014-05-14 17:21:05] now we support up to tlsv1.2, iirc\\ [2014-05-14 17:21:16] tls1.3 is out? anyone knows....\\ [2014-05-14 17:21:44] eZz: as we are on security, you said something about two new modules\\ [2014-05-14 17:21:48] can you give more details?\\ [2014-05-14 17:22:56] eZz: are you around? otherwise we came back later to them\\ [2014-05-14 17:23:12] anything else we should do from security point of view?\\ [2014-05-14 17:23:38] it was a proposal with an alerts mailing list?!? or I am mixing it with something else...\\ [2014-05-14 17:24:53] miconda: it will be great for the future jessie maintenance a way to know security related problems\\ [2014-05-14 17:25:37] linuxmaniac: ok\\ [2014-05-14 17:25:50] perhaps we can discuss it on mailing list to get the others involved\\ [2014-05-14 17:25:56] +1\\ ===== continuous integration ===== [2014-05-14 17:26:14] \o/\\ [2014-05-14 17:26:19] and now it’s linuxmaniac on stage\\ [2014-05-14 17:26:28] mika... come to rescue me!\\ [2014-05-14 17:26:57] It will be great to start a discussion about it\\ [2014-05-14 17:27:06] linuxmaniac: :)\\ [2014-05-14 17:27:22] what are our plans, wishes and manpower\\ [2014-05-14 17:27:52] from my part\\ [2014-05-14 17:28:21] I've created the initial jobs for static analysis using cppcheck at jenkins\\ [2014-05-14 17:28:42] linuxmaniac: cppcheck on jenkins shows 0\\ [2014-05-14 17:28:46] when i checked\\ [2014-05-14 17:28:46] Hello to all. Don't know if anyone can hel me.\\ [2014-05-14 17:28:59] miconda: the are two kinds of jobs\\ [2014-05-14 17:29:00] was it run?\\ [2014-05-14 17:29:26] we use Kamailio in our voip network and now we're facing an interoperability problem with a voip provider.\\ [2014-05-14 17:29:32] cppcheck-full is a complete run of everything\\ [2014-05-14 17:29:51] it requires support to PRACK while out voip switch doesn't support it.\\ [2014-05-14 17:29:53] but cppcheck is just a check for the files changed in last commit\\ [2014-05-14 17:30:14] s/last commit/from the last commit checked/\\ [2014-05-14 17:30:24] linuxmaniac: ok\\ [2014-05-14 17:30:38] the full check is quite heavy\\ [2014-05-14 17:30:44] Would it be possible to configure kamailio to generate the PRACK requests and OK to PRACKs?\\ [2014-05-14 17:30:46] like more than 8 hours\\ [2014-05-14 17:30:46] so are there thing reported on all source?\\ [2014-05-14 17:31:02] s/thing/things/\\ [2014-05-14 17:31:18] https://kamailio.sipwise.com/view/cppcheck/job/kamailio40-cppcheck-full/1/cppcheckResult/ \\ [2014-05-14 17:31:33] https://kamailio.sipwise.com/view/cppcheck/job/kamailio41-cppcheck-full/2/cppcheckResult/ \\ [2014-05-14 17:31:54] I didn't have time to check the results yet\\ [2014-05-14 17:32:20] ok\\ [2014-05-14 17:32:21] but Alex did a fast check on some and there are some stupid withs\\ [2014-05-14 17:32:24] looking over them soon\\ [2014-05-14 17:32:25] things\\ [2014-05-14 17:32:41] btw, you wanted to send allerts to mailing list\\ [2014-05-14 17:32:42] for sure there are some false positives\\ [2014-05-14 17:32:44] that would be good\\ [2014-05-14 17:33:05] let me know the email address to be used for sending alerts and I can allow it to sr-dev\\ [2014-05-14 17:33:05] miconda: yes, I would like to send mails when a job fails\\ [2014-05-14 17:33:22] mika: ^\\ [2014-05-14 17:33:33] miconda: sorry, I was afk, now here\\ [2014-05-14 17:33:35] miconda: ok\\ [2014-05-14 17:33:51] yeah, np to notify mailing list, i just need a mail adr :)\\ [2014-05-14 17:34:07] checking what we have as outgoing, a sec\\ [2014-05-14 17:34:20] I plan to trigger the full jobs once a mouth\\ [2014-05-14 17:34:23] mounth\\ [2014-05-14 17:34:26] miconda: please tell me later, when I can tell something about those modules\\ [2014-05-14 17:35:15] mika: it should send mails to sr-dev@lists.sip-router.org\\ [2014-05-14 17:35:29] miconda: correct?\\ [2014-05-14 17:36:02] So, static analysis done, ok?\\ [2014-05-14 17:36:15] yes, to ser-dev@…\\ [2014-05-14 17:36:21] sr-dev@…\\ [2014-05-14 17:36:24] alright\\ [2014-05-14 17:36:46] linuxmaniac: ok\\ [2014-05-14 17:36:49] BTW, If you detect false positives, please report them to me in order to filter them\\ [2014-05-14 17:37:07] next, functional testing\\ [2014-05-14 17:37:13] for valgrind, kamailio can be compiled with system malloc\\ [2014-05-14 17:37:23] insted of pkg_malloc/pkg_free\\ [2014-05-14 17:37:50] miconda: but for shm?\\ [2014-05-14 17:38:07] I think valgrind needs to be teached\\ [2014-05-14 17:38:28] of our *_malloc *_free functions\\ [2014-05-14 17:38:53] doesn't it?\\ [2014-05-14 17:39:01] don’t know how\\ [2014-05-14 17:39:05] could be possible\\ [2014-05-14 17:39:24] I was thinking to try to replace shm_malloc with pkg malloc\\ [2014-05-14 17:39:47] miconda: http://valgrind.org/docs/manual/mc-manual.html#mc-manual.mempools \\ [2014-05-14 17:40:05] there is an API to teach valgrind of custom allocators\\ [2014-05-14 17:40:15] ok\\ [2014-05-14 17:40:19] that might be better\\ [2014-05-14 17:40:30] if we can do that...\\ [2014-05-14 17:40:40] we can test all the memory allocators\\ [2014-05-14 17:41:04] because replacing shm with system malloc might work only for static analysis, definitely not runtime\\ [2014-05-14 17:41:26] I was aware of the API\\ [2014-05-14 17:41:47] so that's why I'm sure we can do this\\ [2014-05-14 17:41:52] ok\\ [2014-05-14 17:42:17] so, now: functional testing …\\ [2014-05-14 17:42:22] I tried myself to figure out how to do it, but failed\\ [2014-05-14 17:42:45] yes, we need some tests in order to be able to do some automatic testing\\ [2014-05-14 17:43:27] is there a framework, or what is your proposal?\\ [2014-05-14 17:43:30] In sipwise I've created some testings with customs scripts\\ [2014-05-14 17:43:43] can the existing test scripts be reused?\\ [2014-05-14 17:43:44] miconda: enabled, mails should be sent from admin@kamailio.sipwise.com\\ [2014-05-14 17:44:32] miconda: https://github.com/sipwise/kamailio-config-tests \\ [2014-05-14 17:44:48] this is heavily based in our platform \\ [2014-05-14 17:45:11] mika: ok, I will allow the address\\ [2014-05-14 17:45:16] But I would like to discuss something generic\\ [2014-05-14 17:45:23] miconda: alright, just let me know if there are any problems\\ [2014-05-14 17:45:39] mika: thanks :-)\\ [2014-05-14 17:45:58] linuxmaniac: in current source code we have ‘test/unit/*’\\ [2014-05-14 17:46:05] scripts done by Henning a while ago\\ [2014-05-14 17:46:18] not sure how actual, not much activity with them lately\\ [2014-05-14 17:47:14] oej point me to asterisk testing framework or something like that\\ [2014-05-14 17:47:29] the one from sipwise uses sipp?\\ [2014-05-14 17:47:36] miconda: yes\\ [2014-05-14 17:48:03] and is any framework there\\ [2014-05-14 17:48:07] the idea is: set the subscribers, settings, etc... \\ [2014-05-14 17:48:10] or something developed inhouse?\\ [2014-05-14 17:48:18] this is done calling our API\\ [2014-05-14 17:48:31] there is nice collection of sipp scripts on github\\ [2014-05-14 17:49:01] <_gv_> If it's of interest, in Truphone Labs we use both sipp and pjsua, triggered by Jenkins. pjsua has a nice interface to manage this\\ [2014-05-14 17:49:12] and then fire some sipp processes, parse the kamailio log, and test the results\\ [2014-05-14 17:49:43] with this we can test the kamailio config flow\\ [2014-05-14 17:49:50] var contents\\ [2014-05-14 17:50:04]