Mobicents weekly dev meeting, October 15, 2008
7 views
Skip to first unread message
Ivelin Ivanov
Oct 15, 2008, 1:56:27 PM10/15/08
to mobicent...@googlegroups.com
Since pidgin lost the log when windows restarted, I am posting the
summary only. Please post the log if you have it.
summary only. Please post the log if you have it.
Summary:
#1 MMS status - pcmu fixes today, code complete, Luis starts
certification process for CR1
#2 All 1.2.GA status - waiting on MMS release. Will run on EAP 4.3.
production,all,default supported
#3 MSS - 0.7 is almost ready. Might make it in 1.2.GA. We need MSS 1.0
on EAP 4.x (before 5.x)
#4 Diameter - Base client and Sh client integration will go in MSS
0.7. Testing with real servers pending
#5 QE- EAP testing, hudson for EAP, MMS next
Eduardo Martins
Oct 15, 2008, 2:44:01 PM10/15/08
to mobicent...@googlegroups.com
The meeting's log:
[16:05] <ivelin> lets get started
[16:05] <ivelin> #1 MMS CR1 status - critical for 1.2.GA
[16:05] <ivelin> #2 ALL 1.2.GA status
[16:06] <ivelin> #3 MSS
[16:06] <ivelin> #4 Diameter
[16:06] <ivelin> #5 QE
[16:06] <ivelin> other?
[16:06] <@alexandre> docs?
[16:07] <ivelin> don't see silas again
[16:07] <ivelin> #6 Docs
[16:07] <abhayani> he sent a update mail
[16:08] <martins> #? JBOSS AS5
[16:09] <ivelin> AS5 after GA is out
[16:09] <ivelin> let's go
[16:10] <ivelin> #1 MMS CR1 status
[16:10] <kulikoff> we are fixing pcmu codec and releasing today
[16:13] <martins> is there something I can do to help?
[16:13] <kulikoff> Eduardo, testing only
[16:13] <ivelin> Amit, your take on the status?
[16:14] <ivelin> Oleg, docs up to date?
[16:14] <kulikoff> yes, Amit updated all docs
[16:14] <abhayani> Till yesterday MMS was having everything it needed
[16:14] <ivelin> ok, great
[16:14] <abhayani> of course with few leak problems and event firing
[16:15] <abhayani> but that we could have solved in RC2
[16:15] <kulikoff> no
[16:15] <kulikoff> we should fix it now
[16:15] <ivelin> there was an issue with mutliple plays
[16:15] <ivelin> is that fixed?
[16:15] <kulikoff> else it still present in CR2, etc
[16:15] <ivelin> don't understand
[16:16] <ivelin> cr1 will still have the multiple play problem?
[16:16] <kulikoff> with alaw everything work now ok
[16:16] <abhayani> Ivelin, its fixed
[16:16] <kulikoff> without leeks and other problem
[16:16] <ivelin> ok
[16:16] <abhayani> any way
[16:16] <kulikoff> I guess I missed pcmu in list of supported codec in some components
[16:17] <ivelin> please point out in the release notes the API changes, so people are not caught by surprise when old code breaks
[16:17] <abhayani> Ivelin, assure you that doco is thorough this time
[16:17] <abhayani> the README.txt carries updates too
[16:18] <ivelin> nice
[16:18] <ivelin> ok, next one then
[16:18] <ivelin> oh, Luis, are you testing CR1?
[16:18] <barreiro> I'm not.
[16:19] <ivelin> well then its not going out then
[16:19] <ivelin> every public release needs to be QE certified
[16:19] <abhayani> I will create a internal binary before I tag SVN
[16:19] <barreiro> I will now, if you guys say that is ready to be tested.
[16:20] <ivelin> please certify that all issues closed are indeed working and covered by testsuite
[16:21] <ivelin> Luis, can you please focus on this now. We need MMS out to unblock All 1.2
[16:22] <barreiro> sure, ivelin. I didn't start it before because MMS has been unstable for the past weeks, sometimes unbuildable.
[16:22] <ivelin> hm
[16:22] <baranowb> that was just hudson glitch
[16:22] <baranowb> it checked out half of the commit
[16:23] <abhayani> hudson didn't find the CR1 changes I made to pom.xml
[16:23] <baranowb> impl was checked fine but tests not for instance
[16:23] <abhayani> also JAIN SLEE 1.2.1.GA was not in JBoss Repo
[16:23] <abhayani> and MMS depends on JSLEE 1.2.1.GA
[16:23] <barreiro> ok ok, if you guys say that is fine, then I'll test it.
[16:23] <abhayani> reported to Eduardo and he has fixed it
[16:24] <abhayani> Luis please wait till MMS team gives you a go ahead
[16:24] <barreiro> ok, let us coordinate in one binary.
[16:25] <ivelin> alright, focused effort. Let's get this out the door
[16:26] <vralev> wait MMS depends on JSLEE??
[16:26] <ivelin> #2 ALL 1.2.GA status
[16:26] <abhayani> vralev : the examples
[16:26] <vralev> ok
[16:26] <martins> and the external interfaces
[16:27] <martins> ivelin, nothing to talk about ALL GA I think, just lacks MMS to test a snapshot
[16:27] <martins> and release
[16:27] <abhayani> is ti ALL 1.2.GA or ALL 1.2.1.GA ?
[16:28] <martins> keeping same schem as before it is 1.2.1
[16:28] <abhayani> ok
[16:29] <ivelin> there is no such thing as 1.2.1GA yet
[16:29] <ivelin> we need 1.2.GA out
[16:29] <martins> what amit means is that is have slee 1.2.1.GA out
[16:29] <abhayani> ivelin: eduardo released JSLEE 1.2.1.GA which has SIP changes
[16:29] <abhayani> and AFIK we are now suppose to use 1.2.1.GA SLEE
[16:29] <martins> so we keep same relation between slee and all versions
[16:30] <martins> so *if* we keep
[16:30] <martins> this GA will be 1.2.1, not 1.2.0
[16:30] <jeand> on a side note
[16:30] <ivelin> ok, JSLEE 1.2.1.GA is fine
[16:30] <jeand> we are pretty advanced on the 0.7 version of MSS
[16:30] <ivelin> I was talking about ALL 1.2.GA
[16:30] <jeand> and with quite new interesting features and bug fixes
[16:31] <jeand> so we were wondering if it would make sense to release it for 1.2.GA
[16:31] <jeand> we can talk of later in MSS part of the agenda
[16:31] <martins> you will have to do a minor release anyway right?
[16:31] <martins> due to MMS and sip stack
[16:31] <jeand> yeah for including MMS
[16:32] <martins> so the release process will happen anyway
[16:32] <jeand> indeed
[16:33] <martins> I guess it's your call if the fixes and new features are 0.7
[16:33] <martins> or 0.6.1.
[16:33] <jeand> but more testing will be required
[16:33] <jeand> those new features have been done on trunk not on 0.6.x branch
[16:34] <vralev> i think the testing required for 0.6.1 would be the same as 0.7
[16:34] <jeand> 0.6 has already been tested right ?
[16:34] <martins> how much testing is involved in mss anyway
[16:34] <jeand> only the MMS change will need to be tested
[16:34] <abhayani> jean if you want you can start making changes for new MSC API
[16:35] <abhayani> its already documented
[16:35] <jeand> the current things not implemented in 0.7 are Basic IPBX, JBON Integration and Diameter Support
[16:35] <ivelin> ok, so it really boils down to having MMS out ASAP and then make a call for mss
[16:35] <abhayani> but its not big effort IMHO
[16:35] <vralev> 0.6 yes, but we are not going to use it in GA anyway
[16:35] <martins> not implemented that were in roadmap?
[16:35] <jeand> abhayani: it has never been a huge effort to do those changes with you around MMS guys
[16:35] <jeand> yes
[16:35] <jeand> it has been 2 weeks of dev only as of now
[16:36] <martins> what are the new features
[16:36] <jeand> the 0.7 was planned to go out in 2 weeks
[16:36] <jeand> but since there is the GA coming
[16:36] <jeand> ...
[16:36] <martins> we are already mixing MSS topic
[16:36] <jeand> Concurrency Control
[16:36] <jeand> Congestion Control
[16:36] <jeand> User Guide with DocBook Documentation
[16:36] <jeand> Mid Call Failover Support for Proxy Applications
[16:36] <jeand> More Extensions : PRACK, OPTIONS, INFO
[16:36] <jeand> Various Enhancements and Bug fixes
[16:36] <jeand> * MSS CLustering improvements
[16:36] <jeand> * Incorrect Session Invalidation for SUBSCRIBE/NOTIFY
[16:36] <jeand> * Implementation of 17.8 P-Asserted Identity Security
[16:36] <jeand> * SipURIImpl.getUser() throws NullPointerException instead of returning null
[16:36] <jeand> * SipFactoryImpl.createSipServletRequest(...) uses App. Session key as Call-ID instead of random, unique ID
[16:36] <jeand> you were done on GA anyway no ?
[16:36] <jeand> :-)
[16:36] <martins> y
[16:37] <@alexandre> and I'll try to give my contribute to 0.7 with diameter support ;)
[16:37] <martins> sounds enough for a 0.7
[16:37] <jeand> Alex said that he could get us a basic Diameter support for end of week
[16:37] <jeand> alexandre: posted at the same time :-)
[16:37] <@alexandre> heh
[16:38] <jeand> so it will only be JON Integration and PBX missing
[16:38] <jeand> anyway PBX was planned to be rescheduled
[16:38] <martins> btw, I though 0.x was only till green tck, so when do you guys do a 1.0 release?
[16:38] <martins> 0.x alwasy sounds like something is missing to be a product
[16:39] <jeand> martins: I changed my mind, 1.0 will be when it will be feature complete, more stable, have all HA implemented and Tooling
[16:39] <@alexandre> yeah, that's wise. 1.0 means "stable"
[16:40] <jeand> it misses still some things to be 1.0. TCK is not enough in my opinion
[16:40] <@alexandre> production ready
[16:40] <martins> it will always miss something
[16:41] <jeand> I layed out the roadmap with all that is needed for 1.0
[16:41] <jeand> http://www.mobicents.org/roadmap.html
[16:41] <jeand> bug fixes will be added as they are discovered
[16:41] <@alexandre> that's why "wine" (...) took 14 years to be 1.0 ;)
[16:41] <jeand> but I want at least a good env to develop on MSS
[16:42] <ivelin> agree
[16:42] <martins> you have jboss 5 in 0.8, do you know that jboss cache in AS 5 is not the same as in 4?
[16:42] <jeand> I do
[16:42] <jeand> it is 2.x version
[16:42] <martins> wouldn't be better to do 1.0 in jb 4
[16:42] <ivelin> #4 Diameter
[16:43] <jeand> martins: the supported version in JBCP will be on jb4
[16:43] <jeand> the next version be it 1.0 or post 1.0 will be on jb5
[16:43] <jeand> such as jslee and all mobicents servers
[16:44] <jeand> next supported version I meant
[16:45] <@alexandre> We've been working on some fixes to bad session/activity handling (issues 401 & 402 + 404), wrong identification of Auth/Acc messages (403) and some other minor fixes
[16:45] <martins> we now have a trunk/servers/diameter !!!
[16:45] <@alexandre> hudson made us discover these issues as some tests were failing, still there's one that's failing, still not getting why, it works locally
[16:46] <@alexandre> and yes, we've moved to /servers/
[16:46] <@alexandre> :)
[16:46] <martins> hehe
[16:46] <@alexandre> and now I've just started working on the MSS example so it can go in 0.7
[16:46] <jeand> alexandre: great
[16:46] <martins> one thing
[16:46] <jeand> do it mean that we will feature a HSS server too ?
[16:47] <martins> does that example depends on diameter stuff not present in stack and api bundled with jslee 1.2.1 RA?
[16:47] <@alexandre> jeand: for that we'd need Sh Server, which isn't available yet
[16:48] <@alexandre> it should only depend on base RA. but it needs the mux (which is the SAR)
[16:48] <jeand> alexandre: yeah I meant in the future
[16:48] <@alexandre> jeand: sure :)
[16:48] <jeand> nice
[16:48] <martins> that won't behave well in the ALL (if it goes to 0.7 and 0.7 goes on GA)
[16:49] <martins> on the class loading level
[16:49] <jeand> martins: why the SAR will be shared no ?
[16:49] <@alexandre> at the moment what we need for MSS example is an accounting server simulator, which can handle accounting, manage balance, etc
[16:50] <@alexandre> jeand: yeah, but the SAR is not present in JSLEE 1.2.1.GA
[16:50] <jeand> oh I see
[16:50] <martins> jeand: yes, but the stack in the slee RA is older, as a matter of fact "this" sar is not yet released
[16:50] <jeand> ok we can feature this in 0.8 then
[16:50] <jeand> if it is too problematci regarding class loading
[16:50] <@alexandre> but it can work well on a separate env, I mean, as a MSS exclusive
[16:51] <@alexandre> I'll make it work with MSS, and then we'll see the behavior with JSLEE 1.2.1.GA
[16:51] <martins> hehe how will you see the behavior
[16:52] <martins> siounds complex testing
[16:52] <martins> like undeploy one, deploy the other, etc
[16:52] <jeand> martins: it will not be included by default
[16:52] <jeand> it will be one example that can be tried
[16:52] <martins> only on mss standalone
[16:52] <martins> ?
[16:52] <jeand> yes
[16:52] <@alexandre> martins: well... if both of them can co-exist, that should be good enough, no?
[16:52] <jeand> even standalone
[16:52] <martins> ah cool, that's diff
[16:52] <jeand> it will not be included
[16:52] <martins> alexandre: not about coexist
[16:52] <jeand> it will just be downloadable
[16:53] <jeand> and can be tested out by developpers
[16:53] <martins> it may fail on a specific op due to a specific method not being available
[16:53] <jeand> but not shipped by default
[16:53] <jeand> yet
[16:53] <jeand> only click to call and management console are shipped by default
[16:53] <martins> would be though to guarantee that everything would be working as expected
[16:54] <ivelin> Alex, can you summarize what's the status of Diameter
[16:54] <@alexandre> ok, I'll let you guys know of issues when they arise :)
[16:56] <@alexandre> ivelin: Base with fixed problems, better session handling. Sh-Client working fine, tested with seagull. Moved to /servers/diameter, so it's now a "new" server, and can have independent releases. And now starting MSS integration
[16:57] <martins> sh server after that? :-)
[16:57] <@alexandre> probably :)
[16:57] <ivelin> and you're targeting release with mss integration before mss 0.7 goes out?
[16:57] <@alexandre> yup
[16:58] <@alexandre> it's actually more of a MSS feature than a Diameter
[16:59] <@alexandre> again, as a reminder, this is the kind of integration we are aiming at: http://www.apps.ietf.org/rfc/rfc4006.html#sec-A.2
[17:00] <@alexandre> but probably with a fixed-rate cost instead of time-based
[17:00] <jeand> without the REGISTER part of the call flow
[17:00] <@alexandre> yeah, but that should be easy to add, what can bring more trouble is actually the time-based charging
[17:00] <ivelin> please summarize whats the diff in kind of charging
[17:01] <@alexandre> it's event charging (one call costs X independent on duration) vs unit charging (every minute costs Y)
[17:02] <baranowb> differetn scenarios - sms vs phone call
[17:02] <@alexandre> sort of, but didn't want to put it that way eheheh, cause we're talking of phone calls in both ;)
[17:03] <baranowb> yeah, Ivlein
[17:03] <ivelin> got it
[17:03] <baranowb> it always depends on how user app will charge
[17:03] <ivelin> thank you
[17:03] <vralev> what product can the users use as CC Server?
[17:03] <baranowb> phone call can be event charging
[17:03] <@alexandre> vralev: that's actually the missing part... we can use some 3rd party sw or have our own
[17:04] <baranowb> Marben I think has this
[17:04] <baranowb> Ulticom and trafix maybe
[17:04] <@alexandre> yeah, that's what I'm evaluating at the moment
[17:05] <@alexandre> ericsson sdk provides this, but it's a gui tool, dunno how handy it is to use it.
[17:06] <vralev> would be good if there is something that works on both win and lin
[17:06] <@alexandre> it works, it's java based
[17:06] <vralev> ericson sdk only works on win
[17:07] <@alexandre> not true
[17:07] <@alexandre> I've used it on RHEL guess
[17:08] <vralev> weird
[17:08] <vralev> do you have a link?
[17:09] <@alexandre> http://www.ericsson.com/mobilityworld/sub/open/technologies/charging_solutions/tools/diameter_charging_sdk
[17:09] <baranowb> You have to register and login
[17:09] <@alexandre> http://www.ericsson.com/mobilityworld/developerszonedown/downloads/tools/charging_solutions/ChargingSDK-1_0_D31E.zip
[17:09] <@alexandre> see if it works without login
[17:10] <vralev> yes that's something different than what i downloaded last time
[17:11] <ivelin> lets move on
[17:11] <ivelin> #5 QE
[17:13] <ivelin> Luis?
[17:13] <barreiro> my turn
[17:14] ← abhayani left IRC. ("Leaving.")
[17:15] <barreiro> as agreed last week, I modified mobicents-all and other release scripts so that they can produce JBCP binaries ...
[17:15] <ivelin> what does that mean, remind us
[17:15] <barreiro> it is currently on hudson a job that builds a JBCP binary, based on jboss EAP 4.3 - CP02
[17:16] <barreiro> the latest JBCP binary will be on http://hudson.qa.jboss.com/hudson/view/Mobicents/job/JBCP-Release/ws/
[17:17] <ivelin> yep
[17:18] → abhayani joined the channel.
[17:18] <ivelin> so what's the word, we're going with eap 4.3 or 4.2?
[17:18] <ivelin> there was an issue with WS that Bartek brought up
[17:18] <barreiro> I already spotted a problem with management-console. It seems that some kind of authorization is required for accessing JMX ...
[17:19] <vralev> we are going with EAP 4.3 CP2
[17:19] <baranowb> yeah, jmx console in EAP needs auth
[17:19] <barreiro> ... but I guess it will be the same with both versions. Another thing is the default configuration, because in EAP the default configuration is not default, it's production
[17:19] <vralev> barreiro, there is a noauth version of EAP, should not be secured there, but it's easy to disable wither way
[17:20] <barreiro> ... I can get that version. Ivelin, you have the word.
[17:20] <ivelin> well, there should be only one version of EAP sold to customers
[17:21] <ivelin> what's the no-auth one?
[17:21] <vralev> customers can choose between zip,rpm and installer
[17:21] <barreiro> let my try to put the default config problem again ... when you start jboss with bin/run.sh it will fail because production configuration is removed by the scripts. you have to start it with bin/run.sh -c default
[17:22] <martins> what does "production" default configuration means
[17:22] <baranowb> EAP starts by default
[17:22] <baranowb> in prodction node
[17:22] <vralev> the zip comes in two versions - with and without authentication configured
[17:22] <baranowb> regular jboss starts by default in default node
[17:22] <martins> well, let's just leave the production configuration in the script
[17:23] <martins> that is, delete everything except all, default and production
[17:24] <barreiro> or leave just produtcion and all configurations.
[17:24] <martins> w8, since production is default in EAP, does that means we should use all and production ?
[17:24] <martins> what is the diff between default and production
[17:25] <ivelin> yes, we should populate production as well
[17:25] <abhayani> I think production also have clusering
[17:25] ← baranowb left IRC. (Client closed connection)
[17:25] <ivelin> I think the diff is logging level and tuning params
[17:25] <abhayani> Vladimir am I right?
[17:25] <vralev> production is based on all
[17:25] <vralev> yes
[17:26] <vralev> production is basically all with console logging turned off
[17:26] <abhayani> we can have default and production
[17:26] <martins> I would prefer default, all and production
[17:27] <vralev> actually our clustering is experimental yet
[17:27] <vralev> so we don't really encourage using it in production right?
[17:28] <jeand> vralev: if it works why not :-) ?
[17:29] <vralev> well, there are many limitations
[17:30] <vralev> but I guess the users will want to use the other cluster features from production - jndi,http,etc so may be we should include it just for that
[17:31] <jeand> the clustering is enabled only
[17:32] <jeand> if the developer specify it in its app deployment descriptor
[17:32] <vralev> right nevermind then
[17:33] <ivelin> we should advertise clustering in "JBCP 1.2
[17:33] <ivelin> and improve it as needed
[17:33] <ivelin> JBCP 1.2.1 or some other minor version of 1.2 should have decent clustering
[17:34] <ivelin> its a highly requested feature for about a year now
[17:34] <ivelin> so most customers will expect it
[17:34] <jeand> we currently support uas and proxy type of applications
[17:34] <jeand> in clustering
[17:34] <jeand> not b2bua nor converged apps
[17:34] <ivelin> when is that? 0.8?
[17:35] <jeand> even though b2bua might be working for end of week...
[17:35] <jeand> yeah b2bua was planned for 0.8
[17:35] <jeand> and converged 0.9
[17:35] <barreiro> so, let me ask again. What should be the configurations ??? (all + prod + default) (default + all) (default + prod) (prod + all)
[17:35] <ivelin> default, prod, all
[17:35] <barreiro> ok, thanks.
[17:36] <vralev> btw, i think there is also the problem of putting references to other sessions in session attributes, which some users would attempt
[17:36] <jeand> yeah that might be problematic
[17:38] <ivelin> stupid cases like that can be dealt via documentation
[17:38] <ivelin> telling them not to do this
[17:39] <barreiro> ... can I continue !?
[17:39] <ivelin> Luis, what's the plan for QE other than MMS testing and eventually ALL 1.2 release testing
[17:39] <ivelin> sure
[17:40] <barreiro> ok, I've been struggling to set up TCK tests on JBCP, that will be the next step on that front.
[17:41] <jeand> sorry guys I've got to go
[17:41] <barreiro> ... also, as a background task, I've been setting up a MSS 0.6 instalation on perf lab ..
[17:42] <barreiro> ... right now I'm testing with 250 cps ... and CPU is pretty low. I'll update more on this once I push it to the limit.
[17:42] <jeand> barreiro: we handle 250 cps ?
[17:42] <jeand> what is the hardware ?
[17:43] <barreiro> may be more. The hardware is the same as always ...
[17:43] <jeand> hum I like it :-)
[17:43] <barreiro> 4x Dual-Core AMD Opteron(tm) Processor 2220 SE --- 4GB RAM
[17:44] <vralev> 0.7 will be a bit slower unfortunately :)
[17:44] <jeand> vralev: maybe not in None concurrency control mode
[17:44] <martins> vralev: the stack is faster
[17:45] <vralev> ivelin: using sessions like that is allowed and used in apps like c2c
[17:45] <martins> I get +10-12% on my laptop and slee
[17:45] <barreiro> vralev, maybe not if the readings of CPU that I have are true. We have plenty of CPU cycles to spare.
[17:45] <vralev> martins, i hope that stack will compensate then :)
[17:46] <vralev> jeand: yes for perf we should use None i guess
[17:46] <jeand> yeah that makes sense
[17:46] <jeand> ok really need to go but that's a very good news luis u made my day
[17:46] <jeand> :-)
[17:47] <barreiro> jeand, wait for tomorrow. I'll make your WEEK :D
[17:47] <martins> jeand: do you use jboss cache if clustering is not enabled?
[17:47] <jeand> LOL
[17:47] <jeand> martins: no
[17:47] <martins> oh :( was thinking the low performance of the 1.4 jbc myth was dead
[17:48] <ivelin> ok, thanks, Luis
[17:48] <ivelin> any time for smartfrog?
[17:49] <jeand> ok later guys
[17:49] <jeand> really need to go now :-)
[17:49] <jeand> sorry about thayt
[17:49] ← ivelin left IRC. (Client closed connection)
[17:50] <barreiro> ivelin, not yet. It was completely impossible.
[17:52] <martins> I have to go and pick my babies too
[17:52] <martins> see you all later
[17:53] <martins> or brb if you stay after meeting
[17:53] <martins> :)
[17:54] <@alexandre> cya
[17:55] → ivelin joined the channel.
[17:55] <ivelin> windows sneaked a reboot on me
[17:55] <ivelin> Luis, I couldn't see your reponse on smartfrog
[17:55] ← jeand left IRC. (Ping timeout)
[17:55] <barreiro> ivelin, not yet. It was completely impossible.
[17:56] <barreiro> I trade it for these MSS perf tests ;)
[17:59] <ivelin> ok
[17:59] <ivelin> anything else guys?
[18:00] <ivelin> Eduardo, can you please send notes, pidgin lost the log when Windoze restarted
[18:01] <ivelin> thanks for joining then
[18:01] <ivelin> same time, same place next week
[16:05] <ivelin> lets get started
[16:05] <ivelin> #1 MMS CR1 status - critical for 1.2.GA
[16:05] <ivelin> #2 ALL 1.2.GA status
[16:06] <ivelin> #3 MSS
[16:06] <ivelin> #4 Diameter
[16:06] <ivelin> #5 QE
[16:06] <ivelin> other?
[16:06] <@alexandre> docs?
[16:07] <ivelin> don't see silas again
[16:07] <ivelin> #6 Docs
[16:07] <abhayani> he sent a update mail
[16:08] <martins> #? JBOSS AS5
[16:09] <ivelin> AS5 after GA is out
[16:09] <ivelin> let's go
[16:10] <ivelin> #1 MMS CR1 status
[16:10] <kulikoff> we are fixing pcmu codec and releasing today
[16:13] <martins> is there something I can do to help?
[16:13] <kulikoff> Eduardo, testing only
[16:13] <ivelin> Amit, your take on the status?
[16:14] <ivelin> Oleg, docs up to date?
[16:14] <kulikoff> yes, Amit updated all docs
[16:14] <abhayani> Till yesterday MMS was having everything it needed
[16:14] <ivelin> ok, great
[16:14] <abhayani> of course with few leak problems and event firing
[16:15] <abhayani> but that we could have solved in RC2
[16:15] <kulikoff> no
[16:15] <kulikoff> we should fix it now
[16:15] <ivelin> there was an issue with mutliple plays
[16:15] <ivelin> is that fixed?
[16:15] <kulikoff> else it still present in CR2, etc
[16:15] <ivelin> don't understand
[16:16] <ivelin> cr1 will still have the multiple play problem?
[16:16] <kulikoff> with alaw everything work now ok
[16:16] <abhayani> Ivelin, its fixed
[16:16] <kulikoff> without leeks and other problem
[16:16] <ivelin> ok
[16:16] <abhayani> any way
[16:16] <kulikoff> I guess I missed pcmu in list of supported codec in some components
[16:17] <ivelin> please point out in the release notes the API changes, so people are not caught by surprise when old code breaks
[16:17] <abhayani> Ivelin, assure you that doco is thorough this time
[16:17] <abhayani> the README.txt carries updates too
[16:18] <ivelin> nice
[16:18] <ivelin> ok, next one then
[16:18] <ivelin> oh, Luis, are you testing CR1?
[16:18] <barreiro> I'm not.
[16:19] <ivelin> well then its not going out then
[16:19] <ivelin> every public release needs to be QE certified
[16:19] <abhayani> I will create a internal binary before I tag SVN
[16:19] <barreiro> I will now, if you guys say that is ready to be tested.
[16:20] <ivelin> please certify that all issues closed are indeed working and covered by testsuite
[16:21] <ivelin> Luis, can you please focus on this now. We need MMS out to unblock All 1.2
[16:22] <barreiro> sure, ivelin. I didn't start it before because MMS has been unstable for the past weeks, sometimes unbuildable.
[16:22] <ivelin> hm
[16:22] <baranowb> that was just hudson glitch
[16:22] <baranowb> it checked out half of the commit
[16:23] <abhayani> hudson didn't find the CR1 changes I made to pom.xml
[16:23] <baranowb> impl was checked fine but tests not for instance
[16:23] <abhayani> also JAIN SLEE 1.2.1.GA was not in JBoss Repo
[16:23] <abhayani> and MMS depends on JSLEE 1.2.1.GA
[16:23] <barreiro> ok ok, if you guys say that is fine, then I'll test it.
[16:23] <abhayani> reported to Eduardo and he has fixed it
[16:24] <abhayani> Luis please wait till MMS team gives you a go ahead
[16:24] <barreiro> ok, let us coordinate in one binary.
[16:25] <ivelin> alright, focused effort. Let's get this out the door
[16:26] <vralev> wait MMS depends on JSLEE??
[16:26] <ivelin> #2 ALL 1.2.GA status
[16:26] <abhayani> vralev : the examples
[16:26] <vralev> ok
[16:26] <martins> and the external interfaces
[16:27] <martins> ivelin, nothing to talk about ALL GA I think, just lacks MMS to test a snapshot
[16:27] <martins> and release
[16:27] <abhayani> is ti ALL 1.2.GA or ALL 1.2.1.GA ?
[16:28] <martins> keeping same schem as before it is 1.2.1
[16:28] <abhayani> ok
[16:29] <ivelin> there is no such thing as 1.2.1GA yet
[16:29] <ivelin> we need 1.2.GA out
[16:29] <martins> what amit means is that is have slee 1.2.1.GA out
[16:29] <abhayani> ivelin: eduardo released JSLEE 1.2.1.GA which has SIP changes
[16:29] <abhayani> and AFIK we are now suppose to use 1.2.1.GA SLEE
[16:29] <martins> so we keep same relation between slee and all versions
[16:30] <martins> so *if* we keep
[16:30] <martins> this GA will be 1.2.1, not 1.2.0
[16:30] <jeand> on a side note
[16:30] <ivelin> ok, JSLEE 1.2.1.GA is fine
[16:30] <jeand> we are pretty advanced on the 0.7 version of MSS
[16:30] <ivelin> I was talking about ALL 1.2.GA
[16:30] <jeand> and with quite new interesting features and bug fixes
[16:31] <jeand> so we were wondering if it would make sense to release it for 1.2.GA
[16:31] <jeand> we can talk of later in MSS part of the agenda
[16:31] <martins> you will have to do a minor release anyway right?
[16:31] <martins> due to MMS and sip stack
[16:31] <jeand> yeah for including MMS
[16:32] <martins> so the release process will happen anyway
[16:32] <jeand> indeed
[16:33] <martins> I guess it's your call if the fixes and new features are 0.7
[16:33] <martins> or 0.6.1.
[16:33] <jeand> but more testing will be required
[16:33] <jeand> those new features have been done on trunk not on 0.6.x branch
[16:34] <vralev> i think the testing required for 0.6.1 would be the same as 0.7
[16:34] <jeand> 0.6 has already been tested right ?
[16:34] <martins> how much testing is involved in mss anyway
[16:34] <jeand> only the MMS change will need to be tested
[16:34] <abhayani> jean if you want you can start making changes for new MSC API
[16:35] <abhayani> its already documented
[16:35] <jeand> the current things not implemented in 0.7 are Basic IPBX, JBON Integration and Diameter Support
[16:35] <ivelin> ok, so it really boils down to having MMS out ASAP and then make a call for mss
[16:35] <abhayani> but its not big effort IMHO
[16:35] <vralev> 0.6 yes, but we are not going to use it in GA anyway
[16:35] <martins> not implemented that were in roadmap?
[16:35] <jeand> abhayani: it has never been a huge effort to do those changes with you around MMS guys
[16:35] <jeand> yes
[16:35] <jeand> it has been 2 weeks of dev only as of now
[16:36] <martins> what are the new features
[16:36] <jeand> the 0.7 was planned to go out in 2 weeks
[16:36] <jeand> but since there is the GA coming
[16:36] <jeand> ...
[16:36] <martins> we are already mixing MSS topic
[16:36] <jeand> Concurrency Control
[16:36] <jeand> Congestion Control
[16:36] <jeand> User Guide with DocBook Documentation
[16:36] <jeand> Mid Call Failover Support for Proxy Applications
[16:36] <jeand> More Extensions : PRACK, OPTIONS, INFO
[16:36] <jeand> Various Enhancements and Bug fixes
[16:36] <jeand> * MSS CLustering improvements
[16:36] <jeand> * Incorrect Session Invalidation for SUBSCRIBE/NOTIFY
[16:36] <jeand> * Implementation of 17.8 P-Asserted Identity Security
[16:36] <jeand> * SipURIImpl.getUser() throws NullPointerException instead of returning null
[16:36] <jeand> * SipFactoryImpl.createSipServletRequest(...) uses App. Session key as Call-ID instead of random, unique ID
[16:36] <jeand> you were done on GA anyway no ?
[16:36] <jeand> :-)
[16:36] <martins> y
[16:37] <@alexandre> and I'll try to give my contribute to 0.7 with diameter support ;)
[16:37] <martins> sounds enough for a 0.7
[16:37] <jeand> Alex said that he could get us a basic Diameter support for end of week
[16:37] <jeand> alexandre: posted at the same time :-)
[16:37] <@alexandre> heh
[16:38] <jeand> so it will only be JON Integration and PBX missing
[16:38] <jeand> anyway PBX was planned to be rescheduled
[16:38] <martins> btw, I though 0.x was only till green tck, so when do you guys do a 1.0 release?
[16:38] <martins> 0.x alwasy sounds like something is missing to be a product
[16:39] <jeand> martins: I changed my mind, 1.0 will be when it will be feature complete, more stable, have all HA implemented and Tooling
[16:39] <@alexandre> yeah, that's wise. 1.0 means "stable"
[16:40] <jeand> it misses still some things to be 1.0. TCK is not enough in my opinion
[16:40] <@alexandre> production ready
[16:40] <martins> it will always miss something
[16:41] <jeand> I layed out the roadmap with all that is needed for 1.0
[16:41] <jeand> http://www.mobicents.org/roadmap.html
[16:41] <jeand> bug fixes will be added as they are discovered
[16:41] <@alexandre> that's why "wine" (...) took 14 years to be 1.0 ;)
[16:41] <jeand> but I want at least a good env to develop on MSS
[16:42] <ivelin> agree
[16:42] <martins> you have jboss 5 in 0.8, do you know that jboss cache in AS 5 is not the same as in 4?
[16:42] <jeand> I do
[16:42] <jeand> it is 2.x version
[16:42] <martins> wouldn't be better to do 1.0 in jb 4
[16:42] <ivelin> #4 Diameter
[16:43] <jeand> martins: the supported version in JBCP will be on jb4
[16:43] <jeand> the next version be it 1.0 or post 1.0 will be on jb5
[16:43] <jeand> such as jslee and all mobicents servers
[16:44] <jeand> next supported version I meant
[16:45] <@alexandre> We've been working on some fixes to bad session/activity handling (issues 401 & 402 + 404), wrong identification of Auth/Acc messages (403) and some other minor fixes
[16:45] <martins> we now have a trunk/servers/diameter !!!
[16:45] <@alexandre> hudson made us discover these issues as some tests were failing, still there's one that's failing, still not getting why, it works locally
[16:46] <@alexandre> and yes, we've moved to /servers/
[16:46] <@alexandre> :)
[16:46] <martins> hehe
[16:46] <@alexandre> and now I've just started working on the MSS example so it can go in 0.7
[16:46] <jeand> alexandre: great
[16:46] <martins> one thing
[16:46] <jeand> do it mean that we will feature a HSS server too ?
[16:47] <martins> does that example depends on diameter stuff not present in stack and api bundled with jslee 1.2.1 RA?
[16:47] <@alexandre> jeand: for that we'd need Sh Server, which isn't available yet
[16:48] <@alexandre> it should only depend on base RA. but it needs the mux (which is the SAR)
[16:48] <jeand> alexandre: yeah I meant in the future
[16:48] <@alexandre> jeand: sure :)
[16:48] <jeand> nice
[16:48] <martins> that won't behave well in the ALL (if it goes to 0.7 and 0.7 goes on GA)
[16:49] <martins> on the class loading level
[16:49] <jeand> martins: why the SAR will be shared no ?
[16:49] <@alexandre> at the moment what we need for MSS example is an accounting server simulator, which can handle accounting, manage balance, etc
[16:50] <@alexandre> jeand: yeah, but the SAR is not present in JSLEE 1.2.1.GA
[16:50] <jeand> oh I see
[16:50] <martins> jeand: yes, but the stack in the slee RA is older, as a matter of fact "this" sar is not yet released
[16:50] <jeand> ok we can feature this in 0.8 then
[16:50] <jeand> if it is too problematci regarding class loading
[16:50] <@alexandre> but it can work well on a separate env, I mean, as a MSS exclusive
[16:51] <@alexandre> I'll make it work with MSS, and then we'll see the behavior with JSLEE 1.2.1.GA
[16:51] <martins> hehe how will you see the behavior
[16:52] <martins> siounds complex testing
[16:52] <martins> like undeploy one, deploy the other, etc
[16:52] <jeand> martins: it will not be included by default
[16:52] <jeand> it will be one example that can be tried
[16:52] <martins> only on mss standalone
[16:52] <martins> ?
[16:52] <jeand> yes
[16:52] <@alexandre> martins: well... if both of them can co-exist, that should be good enough, no?
[16:52] <jeand> even standalone
[16:52] <martins> ah cool, that's diff
[16:52] <jeand> it will not be included
[16:52] <martins> alexandre: not about coexist
[16:52] <jeand> it will just be downloadable
[16:53] <jeand> and can be tested out by developpers
[16:53] <martins> it may fail on a specific op due to a specific method not being available
[16:53] <jeand> but not shipped by default
[16:53] <jeand> yet
[16:53] <jeand> only click to call and management console are shipped by default
[16:53] <martins> would be though to guarantee that everything would be working as expected
[16:54] <ivelin> Alex, can you summarize what's the status of Diameter
[16:54] <@alexandre> ok, I'll let you guys know of issues when they arise :)
[16:56] <@alexandre> ivelin: Base with fixed problems, better session handling. Sh-Client working fine, tested with seagull. Moved to /servers/diameter, so it's now a "new" server, and can have independent releases. And now starting MSS integration
[16:57] <martins> sh server after that? :-)
[16:57] <@alexandre> probably :)
[16:57] <ivelin> and you're targeting release with mss integration before mss 0.7 goes out?
[16:57] <@alexandre> yup
[16:58] <@alexandre> it's actually more of a MSS feature than a Diameter
[16:59] <@alexandre> again, as a reminder, this is the kind of integration we are aiming at: http://www.apps.ietf.org/rfc/rfc4006.html#sec-A.2
[17:00] <@alexandre> but probably with a fixed-rate cost instead of time-based
[17:00] <jeand> without the REGISTER part of the call flow
[17:00] <@alexandre> yeah, but that should be easy to add, what can bring more trouble is actually the time-based charging
[17:00] <ivelin> please summarize whats the diff in kind of charging
[17:01] <@alexandre> it's event charging (one call costs X independent on duration) vs unit charging (every minute costs Y)
[17:02] <baranowb> differetn scenarios - sms vs phone call
[17:02] <@alexandre> sort of, but didn't want to put it that way eheheh, cause we're talking of phone calls in both ;)
[17:03] <baranowb> yeah, Ivlein
[17:03] <ivelin> got it
[17:03] <baranowb> it always depends on how user app will charge
[17:03] <ivelin> thank you
[17:03] <vralev> what product can the users use as CC Server?
[17:03] <baranowb> phone call can be event charging
[17:03] <@alexandre> vralev: that's actually the missing part... we can use some 3rd party sw or have our own
[17:04] <baranowb> Marben I think has this
[17:04] <baranowb> Ulticom and trafix maybe
[17:04] <@alexandre> yeah, that's what I'm evaluating at the moment
[17:05] <@alexandre> ericsson sdk provides this, but it's a gui tool, dunno how handy it is to use it.
[17:06] <vralev> would be good if there is something that works on both win and lin
[17:06] <@alexandre> it works, it's java based
[17:06] <vralev> ericson sdk only works on win
[17:07] <@alexandre> not true
[17:07] <@alexandre> I've used it on RHEL guess
[17:08] <vralev> weird
[17:08] <vralev> do you have a link?
[17:09] <@alexandre> http://www.ericsson.com/mobilityworld/sub/open/technologies/charging_solutions/tools/diameter_charging_sdk
[17:09] <baranowb> You have to register and login
[17:09] <@alexandre> http://www.ericsson.com/mobilityworld/developerszonedown/downloads/tools/charging_solutions/ChargingSDK-1_0_D31E.zip
[17:09] <@alexandre> see if it works without login
[17:10] <vralev> yes that's something different than what i downloaded last time
[17:11] <ivelin> lets move on
[17:11] <ivelin> #5 QE
[17:13] <ivelin> Luis?
[17:13] <barreiro> my turn
[17:14] ← abhayani left IRC. ("Leaving.")
[17:15] <barreiro> as agreed last week, I modified mobicents-all and other release scripts so that they can produce JBCP binaries ...
[17:15] <ivelin> what does that mean, remind us
[17:15] <barreiro> it is currently on hudson a job that builds a JBCP binary, based on jboss EAP 4.3 - CP02
[17:16] <barreiro> the latest JBCP binary will be on http://hudson.qa.jboss.com/hudson/view/Mobicents/job/JBCP-Release/ws/
[17:17] <ivelin> yep
[17:18] → abhayani joined the channel.
[17:18] <ivelin> so what's the word, we're going with eap 4.3 or 4.2?
[17:18] <ivelin> there was an issue with WS that Bartek brought up
[17:18] <barreiro> I already spotted a problem with management-console. It seems that some kind of authorization is required for accessing JMX ...
[17:19] <vralev> we are going with EAP 4.3 CP2
[17:19] <baranowb> yeah, jmx console in EAP needs auth
[17:19] <barreiro> ... but I guess it will be the same with both versions. Another thing is the default configuration, because in EAP the default configuration is not default, it's production
[17:19] <vralev> barreiro, there is a noauth version of EAP, should not be secured there, but it's easy to disable wither way
[17:20] <barreiro> ... I can get that version. Ivelin, you have the word.
[17:20] <ivelin> well, there should be only one version of EAP sold to customers
[17:21] <ivelin> what's the no-auth one?
[17:21] <vralev> customers can choose between zip,rpm and installer
[17:21] <barreiro> let my try to put the default config problem again ... when you start jboss with bin/run.sh it will fail because production configuration is removed by the scripts. you have to start it with bin/run.sh -c default
[17:22] <martins> what does "production" default configuration means
[17:22] <baranowb> EAP starts by default
[17:22] <baranowb> in prodction node
[17:22] <vralev> the zip comes in two versions - with and without authentication configured
[17:22] <baranowb> regular jboss starts by default in default node
[17:22] <martins> well, let's just leave the production configuration in the script
[17:23] <martins> that is, delete everything except all, default and production
[17:24] <barreiro> or leave just produtcion and all configurations.
[17:24] <martins> w8, since production is default in EAP, does that means we should use all and production ?
[17:24] <martins> what is the diff between default and production
[17:25] <ivelin> yes, we should populate production as well
[17:25] <abhayani> I think production also have clusering
[17:25] ← baranowb left IRC. (Client closed connection)
[17:25] <ivelin> I think the diff is logging level and tuning params
[17:25] <abhayani> Vladimir am I right?
[17:25] <vralev> production is based on all
[17:25] <vralev> yes
[17:26] <vralev> production is basically all with console logging turned off
[17:26] <abhayani> we can have default and production
[17:26] <martins> I would prefer default, all and production
[17:27] <vralev> actually our clustering is experimental yet
[17:27] <vralev> so we don't really encourage using it in production right?
[17:28] <jeand> vralev: if it works why not :-) ?
[17:29] <vralev> well, there are many limitations
[17:30] <vralev> but I guess the users will want to use the other cluster features from production - jndi,http,etc so may be we should include it just for that
[17:31] <jeand> the clustering is enabled only
[17:32] <jeand> if the developer specify it in its app deployment descriptor
[17:32] <vralev> right nevermind then
[17:33] <ivelin> we should advertise clustering in "JBCP 1.2
[17:33] <ivelin> and improve it as needed
[17:33] <ivelin> JBCP 1.2.1 or some other minor version of 1.2 should have decent clustering
[17:34] <ivelin> its a highly requested feature for about a year now
[17:34] <ivelin> so most customers will expect it
[17:34] <jeand> we currently support uas and proxy type of applications
[17:34] <jeand> in clustering
[17:34] <jeand> not b2bua nor converged apps
[17:34] <ivelin> when is that? 0.8?
[17:35] <jeand> even though b2bua might be working for end of week...
[17:35] <jeand> yeah b2bua was planned for 0.8
[17:35] <jeand> and converged 0.9
[17:35] <barreiro> so, let me ask again. What should be the configurations ??? (all + prod + default) (default + all) (default + prod) (prod + all)
[17:35] <ivelin> default, prod, all
[17:35] <barreiro> ok, thanks.
[17:36] <vralev> btw, i think there is also the problem of putting references to other sessions in session attributes, which some users would attempt
[17:36] <jeand> yeah that might be problematic
[17:38] <ivelin> stupid cases like that can be dealt via documentation
[17:38] <ivelin> telling them not to do this
[17:39] <barreiro> ... can I continue !?
[17:39] <ivelin> Luis, what's the plan for QE other than MMS testing and eventually ALL 1.2 release testing
[17:39] <ivelin> sure
[17:40] <barreiro> ok, I've been struggling to set up TCK tests on JBCP, that will be the next step on that front.
[17:41] <jeand> sorry guys I've got to go
[17:41] <barreiro> ... also, as a background task, I've been setting up a MSS 0.6 instalation on perf lab ..
[17:42] <barreiro> ... right now I'm testing with 250 cps ... and CPU is pretty low. I'll update more on this once I push it to the limit.
[17:42] <jeand> barreiro: we handle 250 cps ?
[17:42] <jeand> what is the hardware ?
[17:43] <barreiro> may be more. The hardware is the same as always ...
[17:43] <jeand> hum I like it :-)
[17:43] <barreiro> 4x Dual-Core AMD Opteron(tm) Processor 2220 SE --- 4GB RAM
[17:44] <vralev> 0.7 will be a bit slower unfortunately :)
[17:44] <jeand> vralev: maybe not in None concurrency control mode
[17:44] <martins> vralev: the stack is faster
[17:45] <vralev> ivelin: using sessions like that is allowed and used in apps like c2c
[17:45] <martins> I get +10-12% on my laptop and slee
[17:45] <barreiro> vralev, maybe not if the readings of CPU that I have are true. We have plenty of CPU cycles to spare.
[17:45] <vralev> martins, i hope that stack will compensate then :)
[17:46] <vralev> jeand: yes for perf we should use None i guess
[17:46] <jeand> yeah that makes sense
[17:46] <jeand> ok really need to go but that's a very good news luis u made my day
[17:46] <jeand> :-)
[17:47] <barreiro> jeand, wait for tomorrow. I'll make your WEEK :D
[17:47] <martins> jeand: do you use jboss cache if clustering is not enabled?
[17:47] <jeand> LOL
[17:47] <jeand> martins: no
[17:47] <martins> oh :( was thinking the low performance of the 1.4 jbc myth was dead
[17:48] <ivelin> ok, thanks, Luis
[17:48] <ivelin> any time for smartfrog?
[17:49] <jeand> ok later guys
[17:49] <jeand> really need to go now :-)
[17:49] <jeand> sorry about thayt
[17:49] ← ivelin left IRC. (Client closed connection)
[17:50] <barreiro> ivelin, not yet. It was completely impossible.
[17:52] <martins> I have to go and pick my babies too
[17:52] <martins> see you all later
[17:53] <martins> or brb if you stay after meeting
[17:53] <martins> :)
[17:54] <@alexandre> cya
[17:55] → ivelin joined the channel.
[17:55] <ivelin> windows sneaked a reboot on me
[17:55] <ivelin> Luis, I couldn't see your reponse on smartfrog
[17:55] ← jeand left IRC. (Ping timeout)
[17:55] <barreiro> ivelin, not yet. It was completely impossible.
[17:56] <barreiro> I trade it for these MSS perf tests ;)
[17:59] <ivelin> ok
[17:59] <ivelin> anything else guys?
[18:00] <ivelin> Eduardo, can you please send notes, pidgin lost the log when Windoze restarted
[18:01] <ivelin> thanks for joining then
[18:01] <ivelin> same time, same place next week
Bartosz Baranowski
Oct 15, 2008, 5:23:55 PM10/15/08
to mobicent...@googlegroups.com
What about jslee 1.1 tasks and time estimation?
--
Bartosz Baranowski
JBoss R & D
==================================
Word of criticism meant to improve is always step forward.
Bartosz Baranowski
JBoss R & D
==================================
Word of criticism meant to improve is always step forward.
Ivelin Ivanov
Oct 15, 2008, 10:20:31 PM10/15/08
to mobicent...@googlegroups.com
Good point, but lets get All 1.2.GA OUT OUT OUT OUT OUT NOW!
Reply all
Reply to author
Forward
0 new messages