Attendees:
Alex, Amit, Bartek, Eduardo, Jean, Luis, Oleg, Pavel, Vladimir, Ivelin
Summary:
#1 JSLEE: 2.3 fixes, wiki update, EclipSLEE updates; for 3.0 - ported
cluster framework to Infinisan
#2 Presence, RCS: No progress
#3 Diameter: added clustering support to all Diamete apps; 1.4.Beta1
release goal Feb 7
#4 SS7: working on SIGTRAN M3UA layer for Beta 7
#5 MMS: Stuck on problem with unreliable scheduling; Vladimir will
allocate time to help out
#6 QE: focus on JBCP
5.1.ER; adding networks stats to perf regression tests
#7 JBCP: 1.2.10 Scheduled release Feb 17;
1.5.ER in QE
#8 MSS: 1.x added RFC 3263 support and E.164 Number Mapping (ENUM),
working on new SNMP MIBs; 2.x CDI integration with Tomcat 7 and AS7
Log:
-------
[10:13] <ivelin> surf away
[10:13] <mart1ns1> I could do that
[10:13] <mart1ns1> :)
[10:13] <mart1ns1> irc://
irc.freenode.net:6667/#1 JSLEE 3
[10:14] <mart1ns1> This week most of the work with respect to SLEE was
fixing some open issues of 2.3.0 and EclipSLEE
[10:14] <mart1ns1> all of us
[10:14] <mart1ns1> and docs, in concrete the mobicents site and wiki
at
jboss.org [10:14] == vralev [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
joined #mobicents
[10:15] <mart1ns1> in the last 2 days I also started moving the old wiki content
[10:15] <mart1ns1> and took the chance to improve our maven plugins
[10:15] <mart1ns1> wrt 3.x I updated the cluster framework for
Infinispan 5 ALPHA2
[10:15] <mart1ns1> and started on plugin it to the SLEE core
[10:16] <ivelin> infinispan on AS7?
[10:16] <mart1ns1> no, still using that AS5 default profile for the tests
[10:16] <mart1ns1> it's still not plugged into AS7
[10:16] <mart1ns1> Infinispan I mean
[10:18] <mart1ns1> till that is available we are going to use AS5 for
the developments
[10:18] <mart1ns1> also, hopefully still this week the Annotations wiki is set
[10:18] <mart1ns1> for community discussion
[10:19] <ivelin> ok
[10:20] <ivelin> #2
[10:20] <mart1ns1> didn't find time to work on it
[10:21] <mart1ns1> was supposed to do it last thursday and friday, but
the site thing took my full attention
[10:22] <ivelin> #3
[10:23] <alexandrem> we've been working on the 1.4.0.BETA1 issues
[10:23] <alexandrem> cluster improvement is being applied to the
remaining applications we support
[10:24] <alexandrem> some improvement and fine-grained configuration
was added to statistics, allowing to enable/disable single stats
instead of the whole thing
[10:25] <alexandrem> also working on customizing the dictionary to a
better form, the one we have been using is a bit "unnatural"
[10:25] <alexandrem> after that we plan on a few more changes, and
release on 7th Feb, we hope
[10:26] <ivelin> what clustering tests are in-place now?
[10:27] <alexandrem> we have our Charging Server working in cluster
mode and with that we experiment the changes we are performing
[10:28] <alexandrem> but this is not automated, ie, ready to run at hudson
[10:28] <alexandrem> we will do that soon
[10:29] <ivelin> when are you going to set aside time to do the type
of cluster functional testing and load testing that is in place for
SIP?
[10:31] <alexandrem> it's the next step for cluster
[10:31] <alexandrem> we do have improved on perf and on network
traffic, so now we can do that, after the release we can start working
on it
[10:31] <alexandrem> and plug with SIP as well, as discussed last week
[10:33] <ivelin> ok, then one of the main goals for 1.4.Beta2 will be
a cluster test framework on Hudson. Is that possible?
[10:33] <baranowb> y
[10:35] <alexandrem> sure
[10:36] <ivelin> #4 SS7
[10:36] <abhayani> Last week spent on development of stack for M3UA
[10:36] <abhayani> still continuing on FSM, communication layer and
defining APi for MTP3 User
[10:37] <abhayani> for release Beta7 we have decided to stick to TCP
as communication layer instead of SCTP as discussed last week
[10:38] <abhayani> this is to have wide acceptability of M3UA stack
for Beta7 and after we have some feedback we can move to SCTP
[10:38] <abhayani> hopefully by then JDK 7 will be more accepted by community
[10:39] <abhayani> the good news is Dialogic has agreed to contribute cards
[10:39] <abhayani> hopefully we will have it soon, and do some hands
on and ship the Dialogic native driver too with Beta7 binary
[10:40] <ivelin> would the interface with the dialogic carsd be over tcp?
[10:40] <abhayani> irrespective of ss7 hardware used, there are two ways
[10:41] <abhayani> 1) JAIN SLEE + Hardware on same machine ... no IP
involved here
[10:41] <abhayani> 2) SS7 card on remote machine and interacts with
JSLEE over IP using M3UA protocol
[10:41] <abhayani> we are making all these configurable via CLI
[10:42] <abhayani> for point 2) above, the user can define the
Linkset/Links for underlying hardware and then define it as Signaling
gateway specifying the IP : Port for IP
[10:43] <abhayani> the JSLEE then connects to this Signalling Gateway.
[10:43] <abhayani> Bartek is working on ISUP Stack deve. Bartek?
[10:43] <baranowb> waiting for review :)
[10:43] <abhayani> ach .. will do it asap
[10:44] <baranowb> simplified stack and adding missing parameters/messages
[10:44] <ivelin> so how would we connect to dialogic
[10:44] <baranowb> to cover what is needed to setup call
[10:44] <baranowb> than simple RA
[10:44] <baranowb> in plan
[10:47] <ivelin> hello?!
[10:47] <ivelin> is the connection to dialogic via IP/Sigtran or native?
[10:48] <abhayani> Sigtran
[10:48] <oleg__> native
[10:48] <abhayani> lol
[10:48] <abhayani> the crad always connected via native API
[10:48] <oleg__> the code for driver on svn
[10:49] <oleg__> Bartek's freind patches the pom for building it on windows
[10:49] <abhayani> however if we consider scenario 2) above the
SIGTRAN is involved in between JSLEE and Signalling Gateway
[10:50] <ivelin> guys, you are confusing the heck out of me
[10:50] <ivelin> is the RA using a Java SIGTRAN stack to connect to an
IP port on the Dialogic card?
[10:51] <oleg__> ok, see
[10:51] <oleg__> Dialogic has own native Distributed signaling Interface (DSI)
[10:52] <oleg__> DSI consists of modules and there is API defined to
access the modules
[10:53] <ivelin> via IP or native drivers? Or both?
[10:54] <abhayani> In simple words cards always accessed via native
drivers. But byte[] extracted can be used locally if card is local to
JAIN SLEE or access via IP if JSLEE is remote
[10:55] == oleg___ [55ad5769@gateway/web/freenode/ip.85.173.87.105]
has joined #mobicents
[10:55] <oleg___> sorry was disconnected
[10:55] <oleg___> what was the last statement?
[10:55] <abhayani> (10:22:11 IST) oleg__: DSI consists of modules and
there is API defined to access the modules
[10:55] <abhayani> (10:23:16 IST) ivelin: via IP or native drivers? Or both?
[10:55] <abhayani> (10:24:38 IST) abhayani: In simple words cards
always accessed via native drivers. But byte[] extracted can be used
locally if card is local to JAIN SLEE or access via IP if JSLEE is
remote
[10:56] == oleg__ [55ad5769@gateway/web/freenode/ip.85.173.87.105] has
quit [Ping timeout: 240 seconds]
[10:56] <oleg___> so, modules run either on board or on host
[10:56] <ivelin> why don't we use SIGTRAN even in local scenarios to
simplify things?
[10:57] <oleg___> dialogic has own sigtran implementation wich is free
and it is a module
[10:58] <oleg___> so the Java accesses the SS7 network always via
native driver but backing module can be board or sigtran module
[10:58] <ivelin> with the free Dialogic SIGTRAN stack, we only need
high level RAs ?
[10:58] <oleg___> yes
[10:59] <oleg___> and same if you use card
[10:59] <ivelin> good
[10:59] <oleg___> no difference
aaabhayanialexandrembaranowbbarreiroivelinjeanjeffprestesmart1ns1oleg___slegrikTilivilpanvralev
OptionsAdd webchat to your siteFeedbackFrequently asked questionsAbout qwebirc
menu
Status #mobicentsX
[JAIN SLEE 2.3 Supersonic and Mobicents SS7 1.0 BETA 6 are out !]
[11:01] == ivelin2 [46707d02@gateway/web/freenode/ip.70.112.125.2] has
joined #mobicents
[11:01] <ivelin2> so if we decide to focus on Dialigic integration, we
would greatly simplify our work on lower level SS7 layers and will be
able to focus on SS7 application level scenarios?
[11:02] <ivelin2> (my IRC connection dropped)
[11:02] == ivelin [46707d02@gateway/web/freenode/ip.70.112.125.2] has
quit [Ping timeout: 240 seconds]
[11:02] <oleg___> the top layers is a goal
[11:03] == barreiro [~
barreiro@bl22-128-235.dsl.telepac.pt] has quit
[Ping timeout: 245 seconds]
[11:04] <ivelin2> what do you mean, its a goal?
[11:05] <oleg___> the final destination in other word
[11:05] <oleg___> s
[11:06] == barreiro [~
barreiro@bl22-190-217.dsl.telepac.pt] has joined
#mobicents
[11:09] <ivelin2> I am asking from a roadmap perspective.
[11:10] <ivelin2> Do we need to be spending much effort on lower level
APIs, when we can leverage a free SIGTRAN stack for a ss7 card that
seems to work well
[11:10] <abhayani> ivelin the question is do we focus on only Dialogic?
[11:10] <oleg___> The my opinion that SS7 hardware is big block for community
[11:11] <jean> abhayani, what cards people are using ?
[11:11] <jean> I mean typical customer deployments
[11:11] <jean> or would there be a way that they could hook themselves
their own card to our stuff
[11:12] <jean> so that we can focus only on the high level ?
[11:12] <abhayani> jean: this decision mostly driven by service
providers. For larger telcos who has deep pocket, they don't care
[11:12] <oleg___> and people who can contribute code would be benefit
to see chip harware and up side down those guys who can pay for
dialogic won't contribute
[11:12] <abhayani> but there are many small players who would love to
keep cost low
[11:13] <abhayani> and Mobicents should attract ones who are looking
for cheap solution
[11:13] <oleg___> at least there is a risk to spend money for harware
and never start service or not promete it
[11:13] <oleg___> promote*
[11:13] <abhayani> as one who can pay will directly pay to biggies and
get ready made solution
[11:13] <ivelin2> that is true
[11:14] <ivelin2> but we've been working on ss7 for more than a year
[11:14] <ivelin2> and we missed several big opportunities
[11:14] <oleg___> so the starting with chip harware and scale up is a
good strategy for them
[11:14] <ivelin2> because we don't have answers to higher level API questions
[11:15] <ivelin2> so I'm only thinking outloud in terms of priority
[11:15] <ivelin2> wouldn't it be better to have a complete solution,
even if it is initially expensive
[11:15] <abhayani> ivelin2 : not boasting, but SS7 is ocean. Frankly
appart from Oleg, we all had to start from scratch
[11:15] <ivelin2> rather than lose more opportunities while working on
lower level APIs
[11:16] <abhayani> our competition is doing same
[11:16] <abhayani> they are using Ulticom boards
[11:16] <ivelin2> Amit, you are preaching to the chorus. SS7 is still
a major mudball for me.
[11:16] <ivelin2> but let's focus on opportunity cost for a moment
[11:16] <abhayani> and Ulticoms are not cheap either AFAIK
[11:17] <ivelin2> I'm sure Ulticom is not cheap at all
[11:17] <abhayani> 1) so I agree we need to prioritize things 2) make
stable what we have rather than adding more and more
[11:17] <ivelin2> but if we had support for that, we could have closed
additional business
[11:17] <abhayani> trying to follow up with them
[11:18] <abhayani> but looks like they are not very open to this model
[11:18] <abhayani> not sure what are the apprehensive about as they
partner with our competition
[11:19] <oleg___> let's look at OC
[11:19] <abhayani> they replied back saying that they are thinking
about this as Ulticom is not only hardware company but also provide
SS7 s/w layers
[11:19] <abhayani> and can compete with Mobicents
[11:19] <oleg___> they compiled code from ASN and access hardware at
most top layer
[11:20] <oleg___> so if there is no question about price and licence
the shortest way is to buy ASN compiler and implement everything
(except ISUP) in one day
[11:21] <oleg___> I heard (but never tryied) that nokalava is good
http://www.oss.com/ [11:21] <ivelin2> what is the value prop from Okalava for us?
[11:22] <oleg___> they have ASN.1 compiler
[11:23] <oleg___> all specs (except isup) are written on ASN.1 wich
can be compiled to Java
[11:27] <ivelin2> are you saying that we can take the OC approach and
produce the full SS7 APIs and connect them to the Dialogic SIGTRAN
stack?
[11:28] <oleg___> yeah
[11:29] <vralev> once compiled to Java, the Java files themselves can
be made open source under any licence?
[11:30] <ivelin2> it depends on the ASN1 compiler license, but there
are open source asn1 compilers out there
[11:31] <ivelin2> guys, should we just do that then and be done with
this perception that our SS7 support is incomplete?
[11:31] <ivelin2> then we can work on the lower end cards without
sweating it so much
[11:31] == oleg2_ [55ad5769@gateway/web/freenode/ip.85.173.87.105] has
joined #mobicents
[11:32] <oleg2_> Vladimir, I am not sure
[11:32] == oleg___ [55ad5769@gateway/web/freenode/ip.85.173.87.105]
has quit [Ping timeout: 240 seconds]
[11:32] <abhayani> vralev: most probably no
[11:32] <abhayani> ivelin2: I have checked out open source ASN projects
[11:32] <abhayani> most of them are GPL
[11:33] <abhayani> and only one that I liked in them when we satrted
writing our own ASN Java classes
[11:34] <oleg2_> I don't see open source ASN projects in good conditions
[11:34] <oleg2_> most of them at the begining stage and/or for SNMP
[11:34] <oleg2_> or crypto
[11:35] <ivelin2> GPL is fine. It does not imply restrictions on the
generated code
[11:35] <oleg2_> but anyway having trusted ASN compiler will help for
writting test cases
[11:37] <ivelin2> how about an offline thread on this topic. Let's
make some decisions and push ss7 to competitive state.
[11:37] <ivelin2> as it stands, when is the next ss7 release?
[11:37] <abhayani> Feb end
[11:41] <ivelin2> ok next
[11:41] <ivelin2> #5 MMS
[11:42] <oleg2_> Last week I have been working on integration
scheduler with rtp part
[11:43] <oleg2_> Completed the extended test module wich can bu run on Hudson
[11:43] == vralev [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
quit [Ping timeout: 240 seconds]
[11:43] <oleg2_> and gives the whole picture about RTP part: max
performance, losts, failure rates, etc
[11:44] <oleg2_> now I see the problems during "starting" phase and gc
[11:44] == vralev2 [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
joined #mobicents
[11:45] <oleg2_> so the total missrate remains very small (0.01-0.05)
what is very close to limit for software scheduling but the value for
delay is too big
[11:46] <ivelin2> so the TCK tests still fail becaues of GC impact ?
[11:47] <oleg2_> one can happen during job scheduling - the solution
is to have dedicated special job with low priority and subtask
partiotion
[11:47] <oleg2_> Ivalin, did not start TCK at all during that week,
need to solve this problem
[11:50] == oleg__ [55ad5769@gateway/web/freenode/ip.85.173.87.105] has
joined #mobicents
[11:50] <oleg__> GC may delay processing up to 1 second and easy miss tone
[11:51] <oleg__> I tryed to schedule full GC itself and looks like it
has positive results
[11:51] <jean> what do you mean by schedule full GC
[11:51] <jean> ?
[11:51] <oleg__> the total throughput becomes less but GC in this case
more predictable
[11:51] == oleg2_ [55ad5769@gateway/web/freenode/ip.85.173.87.105] has
quit [Ping timeout: 240 seconds]
[11:52] <vralev2> i doubt GC is a problem for the TCK
[11:52] <vralev2> unless you are using really bad JVM options
[11:52] <vralev2> or something leaks
[11:53] <oleg__> Jean, I have job wich can be scheduled as any other
job. The goal is to run GC more frequently so it looks like high
frequency noise
[11:53] <oleg__> Vladimir, as I said GC pauses easy for 1 sec or more
[11:54] <oleg__> the DTMF tone is 50ms
[11:54] <oleg__> so I can miss it 20 times
[11:54] <ivelin2> maybe the TCK can use a fresh pair of eyes
[11:54] <oleg__> especially when player starts before detector
[11:54] <jean> oleg__, calling System.gc ?
[11:55] <oleg__> Jean, yes
[11:55] <oleg__> Ivelin, I am ready but I don't have another pair of eyes
[11:55] <ivelin2> Vladimir would you be able to spend a few days on the MMS TCK?
[11:55] <vralev2> last time i saw 1 sec GC was at 100 SIP cps
[11:56] <oleg__> Ivelin, it does not make sense to continue with TCK
untill I finish the schedler and RTP part
[11:56] <vralev2> ivelin2: I might be busy next few days, but at least
for this particular problem I think I will have the time
[11:56] <oleg__> Vladimir, 100 SIP cps is for kids :)
[11:57] <ivelin2> thanks, Vladimir
[11:57] <oleg__> I asked Vladmir's help for scheduler and he is
awaiting RTP for test
[11:58] <vralev2> yeah I couldnt run the scheduler on my own so once
integrated with RTP it will be much easier for me to review it
[11:58] <ivelin2> I've been stuck on a sticky problem in the past. It
took anyone else to look at it and see the obvious cause. I think
Ranga saved me a few times.
[11:59] <ivelin2> ok, hoping this will be the week of the big breakthrough
[11:59] <ivelin2> #6 QE
[12:01] <barreiro> This week the focus was on JBCP ER
[12:02] <barreiro> MSS has taken longer to test than expected
[12:02] <barreiro> on one side there were some tests that had to be changed
[12:03] <barreiro> and I also had to wait for pavel to be ready to run
the test suite, which in turn showed some errors ...
[12:05] == vralev2 [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
quit [Ping timeout: 240 seconds]
[12:06] <barreiro> so, the conclusion is that in order to be able to
deliver
5.1.0.GA release in a timely manner, we must keep a close eye
on the snapshot releases that are made each weekend
[12:07] <jean> hold on JBCP
5.1.0.GA uses MSS 1.5 right ?
[12:07] <slegrik> jean, right
[12:08] <jean> so why keep an eye on snapshots ?
[12:09] <barreiro> JBCP snapshots and the results of it's tests. That
is what I meant
[12:10] == vralev [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
joined #mobicents
[12:10] <jean> oh ok
[12:10] <barreiro> to make sure that when we make the release we know
exactly what to expect from the tests.
[12:11] <ivelin2> Luis, give me some good news on the performance
regression runs
[12:11] <slegrik> discussed with Luis before, as there were some
issues found during
JBCP-5.1.0.ER release QA, we need to make sure all
those are covered now.. before GA
[12:12] <barreiro> Ok, so besides that I am working on adding the
network stats to one of the HA perf jobs ...
[12:12] <barreiro> I did some research and it is not likely that we
will be able to separate between traffic and replication
[12:12] == Tili [~
Tili@175.110.107.222] has quit [Ping timeout: 260 seconds]
[12:13] <barreiro> ivelin2, not very good ones :(
[12:13] <barreiro> hudson went down on the weekend and monday. I
started the regression tests manually
[12:14] <ivelin2> darn it
[12:14] <ivelin2> this is so annoying
[12:14] <ivelin2> what happened?
[12:15] <barreiro> SLEE passed but MSS failed. Seemed like there was
work in progress on the trunk, so that the call flow could not be
completed
[12:16] <barreiro> the NAS that powers the shared workspace on all
slaves went down. They had to put up a temporary fix, then there were
another outage to fix it permanently.
[12:17] == Tili [~
Tili@175.110.107.222] has joined #mobicents
[12:17] <ivelin2> we will see how permanent it is
[12:17] <barreiro> Once again, I complain to our hudson maintainer,
although I know there is not much he can do.
[12:17] <ivelin2> have you had time to work on roll up reports?
[12:19] == vralev [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
quit [Ping timeout: 240 seconds]
[12:19] == vralev2 [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
joined #mobicents
[12:19] <ivelin2> Luis?
[12:19] <barreiro> I gave some thought on it and I believe I know the
right way to do it, but haven't start any work on that. I just
modified the job configs to keep the results around for more time.
[12:20] <ivelin2> when do you think we'll see it in action?
[12:21] <barreiro> honestly, I don't give it high priority.
[12:24] <ivelin2> ok, but set high priority on running the perf
regression tests.
[12:24] <ivelin2> #7 JBCP 5.1
[12:25] <slegrik>
JBCP-5.1.0.ER status - covered by Luis already
[12:25] <slegrik> Once QA sign off, we will continue to include only
customer issues fixes (plus issues which are reported now during ER
QA) till 23rd of February, which is the code freeze for
JBCP-5.1.0.GA.
[12:25] <slegrik> JBCP-1.2.10 status - there was planned codefreeze
yesterday, but we are waiting for bundle of
[12:25] <slegrik> issues to be verifed from customer side (AFAIK -
talking to Jean, customer will not be able to verify this week).
[12:25] <slegrik> So, we are stacked at this point here.
[12:25] <slegrik> Besides regular Maintenance Patches schedule,there
has been issued 2 requests from customers to prepare HotFixes. One is
above JBCP 1.2.9 and second one
[12:25] <slegrik> is above JBCP-5.0.1. We are waiting for management
decisions on those two requests. In case we will be doing those two,
we need to figure out impact on
[12:25] <slegrik> the JBCP-1.2.10 and JBCP-5.1.0 release dates. It
could mean some slack, still not sure now.
[12:25] <slegrik> This is the currect status so far, besides I have
been updating Productization pages to reflect current
[12:25] <slegrik> process (
http://intranet.corp.redhat.com/ic/intranet/JBCPProductization.html )
and have finished list of 3rd party jars/apis used in JBCP-5.x.
[12:25] <slegrik> (
https://docspace.corp.redhat.com/docs/DOC-55509 )
[12:29] == vralev2 [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
quit [Quit: Page closed]
[12:30] <ivelin2> ok
[12:30] == vralevv [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
joined #mobicents
[12:31] <ivelin2> is there a thread on the vpn about these hot fixes?
[12:31] <ivelin2> when is the scheduled 1.2.10 release date?
[12:31] <slegrik> ivelin2, one of them posted Jean second one I did
forward to you
[12:31] <ivelin2> thanks
[12:32] <slegrik> 17th of February 2011
[12:35] == oleg__ [55ad5769@gateway/web/freenode/ip.85.173.87.105] has
quit [Ping timeout: 240 seconds]
[12:36] <ivelin2> there is no reason why we shouldn't stick to the
regular schedule then
[12:36] <ivelin2> a couple of weeks is not that far
[12:36] <ivelin2> especially when you include packaging, testing and the works
[12:37] <ivelin2> are there still any risks for 5.1 dependencies. Are
all issues worked out. There were some questions about Diameter.
[12:38] <slegrik> ivelin2, I agree - I would go with Maintenance
Patches as schedules as well
[12:39] <slegrik> regarding 5.1 deps, they are all worked out now
[12:40] <slegrik> we need to cover customer issues only and issues
recognized during ER testing
[12:40] <ivelin2> good
[12:40] <ivelin2> #8 MSS 2
[12:40] == vralevv [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
quit [Ping timeout: 240 seconds]
[12:40] == vralev [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
joined #mobicents
[12:40] <ivelin2> Vladimir is having a rough IRC day
[12:41] <vralev> did you get my message?
[12:41] <jean> no
[12:41] <vralev> well yeah i cant follow what's going on here
[12:41] <vralev> so i was just saying there a some calls today about this
[12:42] <vralev> and the other idea is the have a fasterautomated
process for patches so they dont waste time
[12:43] <vralev> * .. to have a faster automated process ...*
[12:43] <vralev> 2-4 hours machine time or so
[12:44] <jean> vralev, only problem is running the perf tests
[12:44] <jean> it always takes time
[12:45] <vralev> like Roger says EAP hotfixes do not run load tests
[12:45] <ivelin2> Vladimir, what is your idea for fast turnaround
[12:45] <vralev> unless the developer evaluated load testing to be a risk
[12:46] <jean> ah didn't get that in the discussion
[12:46] <ivelin2> one idea we have been contemplating is to let
developers sign jars that expire within a certain time frame
[12:46] <vralev> idea is - developer verifies the patch isolated and
passes the jar to machine for testing - TCK + testsuite
aaabhayanialexandrembaranowbbarreiroivelin2jeanjeffprestesmart1ns1slegrikTilivilpanvralev
OptionsAdd webchat to your siteFeedbackFrequently asked questionsAbout qwebirc
menu
Status #mobicentsX
[JAIN SLEE 2.3 Supersonic and Mobicents SS7 1.0 BETA 6 are out !]
[12:48] == ivelin3 [46707d02@gateway/web/freenode/ip.70.112.125.2] has
joined #mobicents
[12:48] <ivelin3> annoying irc
[12:48] <ivelin3> would we need a new tool that takes a new jar and
triggers test run?
[12:48] <vralev> a wget on-liner in hudson will do
[12:49] == ivelin2 [46707d02@gateway/web/freenode/ip.70.112.125.2] has
quit [Ping timeout: 240 seconds]
[12:49] <vralev> only problem is Pavel requires management approval
and needs to wait on someone for signing
[12:50] <vralev> if we can cut these it would also help
[12:50] <jean> we also need to make sure it runs against the correct JBCP tag
[12:50] <jean> and the jar is placed in the correct location
[12:50] <jean> but that ain't much of a burden just manual stuff
[12:50] <jean> or create a branch for the hotfix
[12:50] <jean> commit in the branch
[12:50] <jean> and run the testsuite against that branch
[12:51] <jean> which should be better in line with EAP hot fix creation
[12:52] <ivelin3> ok, so in order to setup this tool, all we need is
ability to sign jars?
[12:53] <jean> for JBCP 5 yes
[12:53] <jean> for JBCP 1.2.x signing is not required and could be
done right away
[12:56] <ivelin3> ok, how can we try it?
aaabhayanialexandrembaranowbbarreiroivelin3jeanjeffprestesmart1ns1slegrikTilivilpanvralev
OptionsAdd webchat to your siteFeedbackFrequently asked questionsAbout qwebirc
[12:57] == ivelin4 [46707d02@gateway/web/freenode/ip.70.112.125.2] has
joined #mobicents
[12:57] <ivelin4> we still must ensure that customers don't run hot
patches even after the CP is out
[12:57] <ivelin4> signed jars helps with that
[12:57] <ivelin4> but if there is no signed jars for 1.2 we can still
end up with a messy deployment
[12:58] <jean> ivelin4, for 1.2 there is no signing since it based on older EAP
[12:58] <jean> only EAP5 introduced signing
[12:59] <slegrik> did you noticed my message that with MEAD we are now
able to produce signed jars directly at build time ?
[12:59] == ivelin3 [46707d02@gateway/web/freenode/ip.70.112.125.2] has
quit [Ping timeout: 240 seconds]
[13:00] <jean> slegrik, oh so if we produce the binary through MEAD
ourselves we could ship the signed jars ?
[13:00] <jean> but that's maybe not ideal
[13:00] <jean> how long does it take ?
[13:00] <slegrik> jean, yeas :) this is what you can do
[13:00] <slegrik> 1/2 an hour
[13:00] <jean> that's a good option
[13:01] <jean> instead of going through the N day process of asking for signing
[13:01] <jean> thx slegrik
[13:01] <slegrik> jean, and dangerous as well - we might end up with mess
[13:01] == vralev [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
quit [Ping timeout: 240 seconds]
[13:02] <slegrik> in case we provide to customers like 10 or more
patches or so, and they apply all - who will then ensure what they do
have installed ?
[13:02] == vralevttt [4f64602e@gateway/web/freenode/ip.79.100.96.46]
has joined #mobicents
[13:02] <jean> yeah that's why I said maybe not ideal
[13:02] <jean> ivelin and vlad are competing for the worst IRC link
[13:02] <jean> :-)
[13:02] <vralevttt> I just had to enter this capcha
http://dl.dropbox.com/u/1379577/Screen%20shot%202011-02-02%20at%208.59.44%20PM.png [13:02] <jean> slegrik, yeah ideally signed jars with a max duration
[13:02] <slegrik> Look at "Signing of jars in the JBCP products / Fas
patches approach for customers" post on jbcp-dev
[13:03] <jean> jars with ttl
[13:03] <jean> james bond jars
[13:04] <slegrik> :) oh my... we will see how it goes
[13:04] <jean> delicate balance of being too agile or not enough agile
[13:04] <vralevttt> just let the developer evaluate the risk is best
[13:05] <slegrik> jean, someone must be the guard... most probably me ;)
[13:05] <vralevttt> multiply by 2 if you dont trust it :)
[13:05] <ivelin4> yes, agree. Developers tend to be too nice, because
they want their baby to succeed.
[13:06] <ivelin4> Pavel, what is the URL to the discussion thread you mentioned
[13:07] <jean> ok regarding MSS 2
[13:07] <jean> CDI integration is on the work on top of Tomcat 7
[13:08] <jean> we will need to derive a custom module
[13:08] == vilpan [~
vilius@88-222-157-136.meganet.lt] has left #mobicents []
[13:08] <jean> for weld to support MSS on tomcat 7 as weld relies on
Tomcat implementation classes
[13:08] <jean> and we override them
[13:08] <slegrik> ivelin4, there is just message I posted on jbcp-dev
with the "Signing of jars in the JBCP products / Fas patches approach
for customers" subject
[13:08] <jean> so we will need to do the same here
[13:09] <jean> we are starting AS7 integration
[13:09] <jean> talking to AS7 team
[13:10] <jean> to see how best integrate into AS7
[13:10] <jean> and override with our custom converged code
[13:10] <jean> handling both HTTP and SIP
[13:10] <jean> on MSS 1.x side
[13:11] <jean> geroges still working on issue 2263 as there was some
loose ends in the patch provided
[13:11] <jean> apart from that RFC 3263 support and E.164 Number
Mapping (ENUM) support has been added
[13:11] <jean> both at MSS level and JAIN SIP Ext level
[13:12] <jean> so that SLEE apps or SIP RA can reuse it easily
[13:12] <jean> next big item is SNMP
[13:13] <jean> and there were customer issues fixes
[13:13] <jean> or rather backporting
[13:13] <jean> both actually
[13:16] <ivelin4> what on SNMP?
[13:18] <ivelin5> I'm setting IRC reset record today
[13:18] <jean>
http://fosdem.org/2011/schedule/event/mobicents2 [13:18] <ivelin5> what on SNMP?
[13:19] <ivelin5> are users asking for specific new SNMP traps?
[13:19] <jean> yes configuration
[13:19] <jean> alarms
[13:19] <jean> agentx
[13:19] <jean> support
[13:19] <jean> specific SIP MIB
[13:19] <jean> something feeling more ready
[13:19] <jean> there is a thread and issue about that
[13:19] == ivelin4 [46707d02@gateway/web/freenode/ip.70.112.125.2] has
quit [Ping timeout: 240 seconds]
[13:20] <jean>
http://code.google.com/p/mobicents/issues/detail?id=1029 [13:20] <jean> thread is
http://groups.google.com/group/mobicents-public/browse_thread/thread/b76125f3dea885de [13:20] == alexandrem [~
unknown@bl14-24-233.dsl.telepac.pt] has quit
[Ping timeout: 250 seconds]
[13:20] == vralevttt [4f64602e@gateway/web/freenode/ip.79.100.96.46]
has quit [Ping timeout: 240 seconds]
[13:22] == vr_ [4f64602e@gateway/web/freenode/ip.79.100.96.46] has
joined #mobicents
[13:22] <ivelin5> ok thanks
[13:22] <ivelin5> what else do we need to cover?
[13:22] <jean> in SNMP ?
[13:23] == alexandrem [~
unknown@bl5-9-254.dsl.telepac.pt] has joined #mobicents
[13:23] <jean> I need to gather all requirements in one place
[13:25] <ivelin5> is the SNMP plugin still shipped in AS7?
[13:25] <jean> we also need to support v3 of the protocol
[13:25] <jean> not yet
[13:25] <jean> but it will be
[13:25] <jean> since the intern jboss guy I talked to
[13:26] <jean> is doing some work on it for being EAP 5.1 and ideally
AS7 compliant
[13:27] <jean> anything else ?
[13:29] <ivelin5> ok, that is it
[13:29] <ivelin5> thanks everyone
[13:29] <ivelin5> good night and hope you don't get hit by power outages
[13:29] <ivelin5> much of Austin is under power outage today