14:10:00 #startmeeting 14:10:00 Meeting started Mon Jun 27 14:10:00 2011 UTC. The chair is pmuir. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:10:00 Useful Commands: #action #agreed #help #info #idea #link #topic. 14:10:29 I'll start unless anyone has a disagreement ;-) 14:11:01 pmuir: lets start 14:11:54 so last week I started on the migration of the user guide to confluence - i hope that some people have managed to take a look at their articles 14:12:15 oops 14:12:18 #topic pmuir 14:12:27 so last week I started on the migration of the user guide to confluence - i hope that some people have managed to take a look at their articles 14:12:29 yes, i did, they look pretty decent 14:12:33 #link so last week I started on the migration of the user guide to confluence - i hope that some people have managed to take a look at their articles 14:12:42 #link https://docs.jboss.org/author/display/ISPN/Home 14:13:06 Yes, the import worked reasonably well... I've had to fix a few bits manually 14:13:28 I've split it up into 4 sections for now - User Guide - fairly obvious 14:13:59 Tutorials - again fairly obvious. At some point I would like to rename this to the getting started guide, and make the User Guide the "Reference Guide" 14:14:06 but I need to juggle some content to do this 14:14:12 then we have FAQs 14:14:35 I decided to put these on confluence rather than the wiki as I think it's a better tool by far for structured content 14:14:51 pmuir: cool, good stuff 14:14:56 then we have videos - these were part of the user guide and I think I should reintegrate them 14:15:13 and finally Developing Infinispan (which needs a rename, but I need help with that 14:15:16 ) 14:15:36 my plan this week is to start going through and getting the style consistent 14:15:44 and to make a list of what we are missing 14:15:59 with that, I will then write up a style guide + start pestering people for content ;-) 14:16:11 so, wdyt about faqs? 14:16:34 and also, what do we call the "Developing Infinispan" guide - Sanne said that Infinispan isn't a framework so I need another word :-) 14:18:19 ok, i guess no one knows ;-) 14:18:28 I will wait for responses on the ML 14:18:32 ok, next one is 14:18:38 re: developing infinispan 14:18:44 yes go on 14:18:50 how about "Information for Infinispan Developers" 14:19:16 Or is that too easly confused with developers who use Infinispan? 14:19:47 Re: the FAQs, I think they should be broken down into sections, but yeah, fine with them being on confluence ratehr than SBS 14:21:30 it's a tricky one, and I think it's caused by the lack of words like "the" etc. in some languages 14:21:52 all I can say is that we found that adding the word framework, library etc. in there helps people get it 14:22:00 first time - which is important 14:22:07 "platform" ? 14:22:12 yeah we could do that 14:22:13 ok 14:22:31 manik: pmuir +1 for platform 14:22:54 #action pmuir rename Developing the Infinispan Framework to Developing the Infinispan Platform 14:23:10 re faqs my plan is to have one faq per page in confluence 14:23:17 and then aggregate them at various levels 14:23:41 what if we use word contributors, i.e "Development guide for Infinispan contributors" 14:24:12 vblagoje: that works too 14:24:20 or just Contributing to Infinispan 14:24:41 I like that 14:24:57 ok, my next item is 14:24:58 #link https://issues.jboss.org/browse/ISPN-1096 14:25:00 jira [3ISPN-1096] Rework application class loading to be more friendly to modular environments [10Open (Unresolved) Feature Request,7 Blocker,6 Pete Muir] https://issues.jboss.org/browse/ISPN-1096 14:25:16 yeah pmuir, that one is the best 14:25:22 I wrote the status and proposal to move this forward on the issue this morning 14:25:28 does anyone disagree? 14:27:35 I'll leave that open and move on to what I am planning 14:27:41 i don't know much about this so I'll refrain from commenting pmuir :-) 14:27:52 this week we have the locking meeting in Edinburgh 14:27:55 pmuir: I haven't had a look at your proposal yet 14:28:03 I will do so later today 14:28:33 manik: basically issue is that all options that I can see for pulling CL into marshalling are fugly without a rearch of marshalling 14:28:41 and I think it's too risky to do that right now 14:29:05 so I propose defering this to 5.1 - esp as user api is in place today, so this would be spi change only 14:29:14 but yeah, we can discuss on the issue 14:29:36 and then otherwise this week I have a load of emails from Manik re docs, faqs etc. to add in to confluence 14:29:41 so that will take me a while 14:30:29 pmuir: ok 14:30:38 i also will start on a plan for "understanding infinispan" to be ready in 2w time - basically just what areas we need to improve on around our collatoral and look for volunteers to help with that 14:30:51 that's me ;-) 14:30:54 who is next? 14:30:56 pmuir: great! 14:30:58 thanks. 14:31:16 ok I'll go next 14:31:19 #topic vblagoje 14:32:00 so last week I spent most of my time working with Bela on a push of JGroups 3.0 towards a first beta release 14:32:30 so we changed a few signatures, most notably start/stop flush and state transfer 14:33:02 wrt to Infinispan 14:33:25 I got really worked up about finding a cause for alleged performance issue in container 14:33:37 vblagoje: yes, I saw your emails there 14:33:44 so I am working on a series of stress tests to verify these 14:33:52 vblagoje: what else do you have open for 5.0? 14:34:28 I have removing flush, which i believe is solved, and enabling those tests that do not run for some reason 14:35:30 in another words ISPN-83 and ISPN-1108 14:35:31 jira [3ISPN-83] Remove dependency on JGroups FLUSH [10Reopened (Unresolved) Task,7 Critical,6 Vladimir Blagojevic] https://issues.jboss.org/browse/ISPN-83 14:35:32 jira [3ISPN-1108] Test(s) ignored by maven [10Open (Unresolved) Bug,7 Blocker,6 Vladimir Blagojevic] https://issues.jboss.org/browse/ISPN-1108 14:36:34 this week I'll focus on DataContainer performance and will resolve ISPN-1108 14:36:36 jira [3ISPN-1108] Test(s) ignored by maven [10Open (Unresolved) Bug,7 Blocker,6 Vladimir Blagojevic] https://issues.jboss.org/browse/ISPN-1108 14:36:48 nothing else on my plate for 5.0 14:38:33 I'll take it from here then 14:38:45 #topic mmarkus 14:38:56 last week I was at Jazoon Tue-Wen 14:39:31 good conference, had lots of chats around infinispan use cases 14:40:04 also there seems to be interest for the stuff I've prototyped a year ago around continuous queries 14:40:55 also met bela, had some beers and discussed the way state transfer is to be implemented - cool stuff 14:41:09 besides that I've tided up my stuff for 5.0 14:41:25 mmarkus: great. Anything still outstanding for 5.0? 14:41:52 manik: afaik I only have a documentation-related JIRA 14:42:05 so nothing really 14:43:18 lets have a look at DataContainer perf. numbers together then :-) 14:43:48 for this week I'll focus on the new locking stuff 14:44:01 vblagoje: yes. that should be interesting 14:44:39 vblagoje: shall we catch up on this on Thursday? Dan will also be here and we can ask him to join us 14:44:42 ok, anything else for mmarkus or shall we move on to manik ? 14:44:51 manik: +1 for "miller lite" beer, they have the best video :) 14:44:56 ok deal mmarkus 14:45:10 pmuir: I'm done 14:45:12 ok, I can speak next 14:45:24 So I was in Rome last week, meeting with Cloud-TM 14:45:45 They have already finished an atomic broadcast based transactional impl for replication and distribution 14:46:02 and as it stands shows no performance improvements when there are no conflicts 14:46:04 #topic manik 14:46:13 but *huge* improvement even with a modest conflict ratio 14:46:18 (tests run on 10 servers) 14:46:33 so very cool stuff, due to the deadlock-free nature of AB 14:46:41 but there are several inefficiencies in their impl 14:46:51 which I will talk about tomorrow/Weds in Edin 14:47:02 which we can advise them on to improve things even further 14:47:15 manik: cool stuff, I'm very curious about it :) 14:47:35 Anyway, I was suggesting that they also refactor their code so we can bring it upstream and have it as a configurable (non-efault) option 14:48:08 manik: does the locking stuff we've been thinking about make sense then? they want to achieve the same result... 14:48:11 THey've unfortunately impl'd their changes by forking infinispan and changing our interceptors to be AB-specific 14:48:17 which breaks our existing schemes 14:48:30 pmuir: ping? 14:48:33 SO they will need to refactor these into new Interceptors (or subclasses of existing ones) 14:48:36 alesj: one min 14:48:42 before we can pull this upstream 14:48:55 And as a part of this work, 14:49:08 they have implemented a TPC-C benchmark access pattern for RadarGun 14:49:23 (This is the high conflict benchmark used by most database vendors) 14:49:43 the next thing they're working on is configuration tuning hints 14:49:56 Basically a manager to monitor performance and log "recommendations" 14:50:08 e.g., buffer sizes, thread pool sizes, timeouts, etc. 14:50:23 manik: I've been granted access to their TPC-C benchmark and I'll take a look at it asap 14:50:33 Their eventual goal is to make this completely self-adjusting but step 1 is to just log recommendations 14:50:40 all in the name of usability. ;) 14:50:52 what is the planned integration release, 5.1? 14:51:00 vblagoje: yes 14:51:13 The rest of this week, apart from the locking/tx meeting in edinburgh (where we will plan integration with CloudTM work) 14:51:38 I'll be in Brussels Sun/Mon representing RHT at the EU panel for cloud-™ funding 14:52:03 Re: 5.0, let me know when you guys are ready for another CR this week and I'll punt one out 14:52:30 i'm ready once we agree on ISPN-1096 14:52:32 jira [3ISPN-1096] Rework application class loading to be more friendly to modular environments [10Open (Unresolved) Feature Request,7 Blocker,6 Pete Muir] https://issues.jboss.org/browse/ISPN-1096 14:52:48 however i did notice the test suite has a lot of problems for me right now 14:52:56 pmuir: regressions? 14:53:28 odd, I don't see it on hudson 14:53:39 actyually yes i do 14:53:54 manik: mind updating the agenda with atomic broadcast stuff? http://bit.ly/kmCFM9 14:55:16 mmarkus: will do 14:55:36 ok, I'll have a look at JIRA and Hudson and will ping you guys later about a release 14:55:36 atm locally I get Test suite progress: tests succeeded: 1146, failed: 15, skipped: 167. 14:55:47 pmuir: wow thats a lot 14:55:52 does anyone else see this? 14:55:59 manik: right ;-) 14:56:16 also i wanted to ask about the status of infinispan on openshift 14:56:23 alesj: ^^^ I know you worked on this? 14:57:18 it must have been a last few commits pmuir because I did not see this last week (thursday) 14:57:52 pmuir: Matej will have a basic cartridge in Flex till next week 14:58:40 vblagoje: manik: maybe it's just me - but someone should try it ;-) 14:58:46 alesj: ok cool thanks 15:01:41 pmuir: I'll give it a go 15:03:55 so we are done with the meeting? 15:04:11 #endmeeting