14:28:44 #startmeeting PowerManagement -- http://fedoraproject.org/wiki/SIGs/PowerManagement 14:28:44 Meeting started Wed Jun 1 14:28:44 2011 UTC. The chair is jskarvad. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:28:44 Useful Commands: #action #agreed #halp #info #idea #link #topic. 14:29:09 hi 14:29:16 hi xkriva11, jmarko 14:29:26 hi 14:29:52 hi 14:30:14 #topic news 14:30:39 I was working on matahari PM agent during last week. 14:30:57 It is almost completed. 14:31:06 The matahari allows you to manage your Linux and Windows machines through AMQP protocol and/or D-BUS 14:31:20 It provides several agents thats exports various methods/properties. 14:31:41 It is especially useful for clusters of Windows/Linux machines. 14:32:12 #link http://fedorahosted.org/matahari/ <-- matahari 14:33:12 the QFM link on the wiki is not working 14:33:27 ehm, QMF 14:34:03 xkriva11: thanks I will forward it to matahari folks 14:34:12 Our PM agent currently allows you to list available PM profiles, get active PM profile and switch the active PM profile. 14:34:27 On Linux it uses tuned for PM profiles manipulation, on Windows it uses PM API. 14:34:49 The matahari interface is same on both platforms, only the PM profile names differs. 14:35:25 In our PM agent implementation the AMQP and D-BUS is currently fully functional on Linux. 14:35:47 The Windows implementation is not yet tested. 14:36:22 The experimental patch that adds the PM agent into current matahari git head is available: 14:36:55 #link http://jskarvad.fedorapeople.org/matahari-pm-agent.patch <-- matahari pm agent experimental patch 14:37:14 In the future we would like to add functionality for measuring power consumption and settings power caps. 14:38:02 Next, short presentation about the PM news in F15 is available: 14:38:29 #link http://jskarvad.fedorapeople.org/F15ReleaseParty_PowerManagement.pdf <-- PM news in F15 14:39:26 Next, PowerTOP was updated to second beta (AKA 1.98) in f15 / rawhide. 14:39:55 It should be in f15 stable now, it contains many bugfixes. 14:40:39 were some issues from the Testing day fixed before the F15 official release? 14:41:46 which issue? 14:42:42 some of the issues reported on the Testing day. I aks only to know if the TD should be organised sooner in the next release 14:44:01 xkriva11: good point 14:44:22 of course if none of them were critical or important... 14:44:40 plese note, we have limited resources 14:45:02 but afaik there shouldn't be critical unresolved bugs now 14:45:31 feel free to provide patches to speed things up, I am going through bugzilla on best effort basis 14:47:07 but planing test day earlier in the devel phase is good idea, I will try for f16 14:48:27 #topic tuned 14:48:27 on the other hand with later term you will get more users who will want to see all the changes in the system 14:49:07 xkriva11: both approaches has pros and cons 14:49:41 I guess that one reason why the last TD had so broad response is that many people wanted to see new Gnome too 14:49:41 xkriva11: in f15 we were limited by feature completeness 14:50:23 xkriva11: it is no point to make TD before the features are completed 14:51:11 I see 14:51:20 For tuned there was spotted problem with the EEEPC SuperHybrid Engine (SHE) support. 14:51:42 The original SHE settings are not restored during tuned profile switching. 14:51:49 The problem is caused by missing cleanup code - I will provide patch for it. 14:52:14 The nasty bug 690177 that is about timeouts during switching of tuned profiles was finally resolved, thanks to mschmidt. 14:52:29 The problem was in conflicting service priorities that caused systemd deadlocks. 14:52:45 If you are experiencing this problem, there is a quick/temporal workaround before the proper fix will be released - yum remove cpuspeed. 14:53:14 No more topic to tuned. 14:54:03 #topic other 14:54:25 We got RFE - the reporter would like to have the alternate default configs for daemons/services that would be optimized for low power consumption. 14:54:46 It is interesting idea but actually it requires a lot of work and measurements. 14:54:55 I will add this request to our wiki. 14:55:01 If there are daemons that impact power consumption then those daemons need fixing 14:55:07 We shouldn't worry about adding support for covering up bugs 14:57:58 mjg59: sure, but I think there is room for tunings, e.g. polling intervals, queue processing, etc 14:58:51 Do you want to test each daemon separatelly 14:59:40 jskarvad: There should be no daemons that poll 15:00:37 Where we do end up with polling daemons (mostly due to legacy horrors), then if it's acceptable to lengthen the timeouts why wouldn't you always do that? 15:03:36 mjg59: I didn't perform any analysis 15:03:54 just forwarding for discussion 15:04:52 * abadger1999 notes fpc meetingwill start soon if there's quorum. 15:05:18 I have no more topic 15:05:42 Any PM related discussion can follow on #fedora-power 15:06:29 thanks all 15:06:36 #endmeeting