**** BEGIN LOGGING AT Wed Mar 20 02:59:58 2013 Mar 20 04:35:41 ever find out what was wrong with maemo.org DocScrutinizer05 ? Mar 20 04:52:43 Macer: sorry? Mar 20 04:57:20 the high load Mar 20 04:57:23 nm Mar 20 05:25:23 ooh, the high load mainly seen on www? Mar 20 05:26:11 it's caused by cronjobs for autobuilder etc, and I think those jobs need a proper refactoring to remove some parallelism Mar 20 05:27:03 it's insane when a job needs >6GB of memory(!) to proceed. And the high load is probbaly just a side effect of this Mar 20 05:29:53 6GB? damn :P Mar 20 05:29:57 I mean, even apt/yum are happy with a few 100MB of RAM to build a complete dependency datastructure of all packages available in a full linux distro. I can't imagine any task in maemo that would need to handle more complex and thus larger datastructures Mar 20 05:30:12 DocScrutinizer05: database preloading? Mar 20 05:30:23 Hurrian: yes, probably Mar 20 05:30:34 TMO is huge. Mar 20 05:30:37 Frickin' huge. Mar 20 05:30:50 tmo is unrelated to www Mar 20 05:31:02 ah Mar 20 05:32:20 it's probably all related to package interface Mar 20 05:32:22 ~pkg Mar 20 05:32:23 hmm... pkg is http://maemo.org/packages/ Mar 20 05:32:39 is the pkg queue that long? Mar 20 05:33:03 and yes, this probably accesses db VM via remote mysql queries done in php jobs on www VM Mar 20 05:34:10 so if each such php-mysql interface loads a complete view of a nonsensical query of db, this will hog RAM for sure, if those jobs run in parallel for several packages Mar 20 05:35:35 php-mysql "SELECT * on maemo_packages" --> damage done Mar 20 05:35:56 I don't think it's THAT bad, but you get the idea Mar 20 05:36:04 it could be that bad. Mar 20 05:36:44 assuming the dev is doing their job properly, the PHP script could simply directly access the column for that package. Mar 20 05:36:57 yup Mar 20 05:37:14 that's why I say the jobs on www need a refactoring Mar 20 05:38:33 this is a very common mistake done by devels accessing db, they are not aware of what actually happens on machine when they do SQL queries etc Mar 20 05:39:31 DocScrutinizer05: god way to setup builider to few 2-4 cores if is 8cores all Mar 20 05:40:10 we have no problems with builder Mar 20 05:40:22 DocScrutinizer05: compile always tale a lots of ram more than installing apps i think... Mar 20 05:40:51 http://213.128.137.6/ganglia/?c=maemo&h=builder1& Mar 20 05:41:33 http://213.128.137.6/ganglia/?c=maemo&h=www& Mar 20 05:43:43 checking the heatmap, www is really getting hammered. Mar 20 05:44:12 oh, and there seems to be load spikes every 20 minutes or so. Mar 20 05:44:20 yes Mar 20 05:44:24 cronjobs Mar 20 05:44:32 every 15 I think Mar 20 05:44:44 and a monster one every hour Mar 20 05:46:16 or rather every 30min Mar 20 05:46:57 it takes 20min to eat up to 6GB of RAM, then collapses Mar 20 05:47:13 DocScrutinizer05: not looks so bad, but drive space is on the edge... anyway when something is build is only one package or few in one time? Mar 20 05:47:47 builder doesn't need drivespace Mar 20 05:48:12 and I think builder is building aone package at a time, from queue Mar 20 05:49:25 in former times we had like 5 or 10 builder engines Mar 20 05:50:13 which again is giving us errors right now, since there's no more user builder2 for example, but scripts occasionally still use that Mar 20 05:50:33 jacekowski is fixing all that Mar 20 05:50:41 during last week or so Mar 20 05:51:31 kudos to jacekowski Mar 20 05:51:57 but there's obviously still a lot left over to cleanup Mar 20 05:52:29 yea when have a ram all take and handle inside ram... Mar 20 05:53:07 whole maemo.org machinery seen massive restructuring when nemein ported it from Nokia infra to their own infra, and nobody ever fixed all that stuff before jacekowski looked into it Mar 20 05:55:50 and we have two problems: understanding how stuff been meant to work before migration#1 (to nemein), and understand and implement how to fix stuff so it work *now* on new infra after migration#2 Mar 20 05:57:15 that the worst... not touching if works when must somthing do after longthat is a hard workaround.... and when moving is added that is really scarry thing Mar 20 05:58:01 it often is a game of "who's faster - you with searching for the existing script that did XY before it got disabled, or I with writing a new script that re-implements same function" Mar 20 05:58:37 documentation on all that: zero Mar 20 05:59:25 usually not even much in header comment of shell/php-scripts Mar 20 06:05:07 documentatikon changed with people working on server because i think all people was own way to maanage the server... Mar 20 06:06:31 for example I asked Nemein 3 days ago how to create new developer accounts. We have no clue, though we guss there must be a script or even a web frontend for that Mar 20 06:14:39 DocScrutinizer05: i have similar when helping to care and develop game server... Mar 20 06:15:48 but that was smal and there is a huge... Mar 20 06:21:00 I fixed some users ssh pubkey on a few VMs, only to learn it gets reset to old after w while. Obviously the master copy of pubkey is stored in db and gets deployed to ~/.ssh/auth*keys every 15min by some unknown cronjob Mar 20 06:29:47 DocScrutinizer05: thats is interesting in db mean mysql? Mar 20 06:34:19 i also have keys in file that contains a list of keys files... ssh keys with keychain is great connection... Mar 20 06:45:49 Drathir: yes, db VM is our central mysql database Mar 20 06:48:03 doc thats amazing idea for me... and also work... great idea... Mar 20 06:48:48 DocScrutinizer05: * Mar 20 06:50:17 big brains can do amazing things... Mar 20 06:52:19 well, without any documentation such things are way more scary and puzzling than amazing Mar 20 06:52:55 whole maemo.org internals is a constant "WTF?!" Mar 20 06:54:11 much of it introduced by midgrad Mar 20 06:54:20 midgard even Mar 20 06:55:21 e.g. we have no idea how to edit content of http://www.maemo.org Mar 20 06:55:47 DocScrutinizer05: bubut the moment when thinking what the hell that works should be interesting... Mar 20 06:56:06 there's a web frontend supposed to be somewhere for that, since midgard is a CMS I've been told. Nobody has a clue where to look for such web frontend though Mar 20 06:57:36 DocScrutinizer05: and looking all files should take a years... :/ Mar 20 07:03:08 i sometimes have that i dont know how i do somethings, but it works like i want... a Mar 20 07:23:31 on a maybe somewhat interesting sidenote: we're running those 45CPU/14hosts on maemo.orh with a total of ~200W Mar 20 07:24:04 who's paying for the electricity? iphh? Mar 20 07:24:08 yep Mar 20 07:24:23 their donation to us Mar 20 07:24:33 <3 Mar 20 07:25:43 *yawn* spam-day on council.m.o ML Mar 20 07:26:27 I guess our spamassassin needs an update Mar 20 07:29:34 DocScrutinizer05: you joke right? Mar 20 07:29:44 ? Mar 20 07:30:19 DocScrutinizer05: with 200W power Mar 20 07:30:27 I don't even know how joge is written Mar 20 07:30:56 PSU1: 98W, PSU2:86W Mar 20 07:31:16 lol Mar 20 07:31:20 according to IPMI Mar 20 07:34:42 I don't even know how to spell "joge" (sorry for my french) Mar 20 07:34:54 DocScrutinizer05: interesting what motherboard there are... Mar 20 07:35:16 http://wiki.maemo.org/Migrating_to_Community-driven_Infrastructure Mar 20 07:36:07 didn't you ask same question a few days ago already? Mar 20 07:36:43 http://wiki.maemo.org/Migrating_to_Community-driven_Infrastructure#Operational_Platform Mar 20 07:46:18 DocScrutinizer05: thats all is perfect organized... Mar 20 07:47:11 i think that a much bigger hard drives... Mar 20 07:51:02 sorry? Mar 20 07:52:33 for all data i think there must be bigger hdd butisnt that... Mar 20 07:55:26 our primary data fits onto 2TB Mar 20 07:56:23 all hdd is less than 3TB on that big community is not bad i think Mar 20 07:58:43 i trying to setup xen in little home server but he bite with qemu :/ Mar 20 07:59:09 btw that ganglia looks little modern that munin... Mar 20 07:59:44 than* Mar 20 08:00:37 WIP = work in progres? Mar 20 08:15:09 DocScrutinizer05: i find a bug on site one core somewhere hide... in section Cpu Cores, RAM (in MB), storage (DISK, in GB), of the VMs Mar 20 08:16:49 well, probably one core is reserved for dom0 Mar 20 08:17:02 WIP is work in progress, yes Mar 20 08:17:18 that table is not exactly up-to-date, it's at least one week behind Mar 20 08:18:06 and i can tell you exactly which job updates authorized_keys Mar 20 08:18:47 :-) Mar 20 08:18:54 */10 * * * * $EXECUTER dav-svn/create_users.php /var/www/homedirs/ Mar 20 08:18:59 i think that small amount of cores to db only 2 ? little suprised me... Mar 20 08:19:24 mysql is pretty fast Mar 20 08:19:29 Drathir: we can juggle CPUs as we need Mar 20 08:19:38 even during runtime Mar 20 08:19:44 you can have 10kq/s on single not so fast core Mar 20 08:19:58 if you have reasonable queries Mar 20 08:20:03 anyways Mar 20 08:20:05 i've got to go Mar 20 08:20:08 yeah! only then Mar 20 08:20:15 cya jacekowski Mar 20 08:24:04 jacekowski: thanks a lot i write that because that really interesting me... Mar 20 08:26:46 DocScrutinizer05: all earlier infrastructure alo was on xen containers ? Mar 20 08:27:05 no Mar 20 08:27:20 probably not, though we have no way to know for sure Mar 20 08:27:51 we had no access to any internals prior to Nemein completing migration#1 Mar 20 08:28:24 hm, how does adding a cpu to a running system even work? :o Mar 20 08:28:52 since originally maemo.org been mixed with other Nokia proprietary stuff on same machines, so they couldn't allow access for us Mar 20 08:29:17 kerio: I guess that's a question for XEN faq Mar 20 08:30:07 mmcblk0: error 1 sending read/write command, response 0x0, card status 0x900 Mar 20 08:30:11 oh shit Mar 20 08:30:16 I'd suspect the virtual CPUs get "hotplugged" Mar 20 08:30:38 ohshit indeed Mar 20 08:32:27 doing a fsck -c Mar 20 08:32:35 on the important partition Mar 20 08:32:52 luckily, i have errors=remount-ro :) Mar 20 08:34:20 DocScrutinizer05: vcpu hotplug is a mess and not really well implemented Mar 20 08:34:35 DocScrutinizer05: and it's one way only Mar 20 08:34:45 what i read the most interestings was xen and openvz system, but dont know if its true... Mar 20 08:35:21 it's just xen Mar 20 08:37:08 nayway, CPU grunt obviously isn't our main concern on maemo.org infra Mar 20 08:37:40 jobs eating 6GB while running... that's a real problem Mar 20 08:40:32 DocScrutinizer05: do you use maemo conversation IM ? Mar 20 08:43:14 DocScrutinizer05: hm, a fsck -c updated the bad blocks list Mar 20 08:43:16 hopefully it's ok now Mar 20 08:44:56 DocScrutinizer05: sharing memor cache if works was be good but when compile i dont know if its possible and second part myst bee running min 2 compiling works to have fenefit of them... Mar 20 08:45:15 benefits* Mar 20 08:46:17 kerio: what files system is used? Mar 20 08:47:18 How can i disable sound/vibra notifications for IMs while i'm in DnD mode ? Mar 20 08:48:01 Also is it possible to initiate skype-video calls from the handset ? Mar 20 08:50:02 Also, there's a 'Advanced Power app' that shows the percentage of the batter, and time to discharge, as i was told about , it provides some 'ticks' that will cause me battery charge, as the cpu does the updates Mar 20 08:50:11 XATRIX: for no sound/vibra choose silent without vibrate. For skype-video just select a contact who can receive a skype video call and there'll be an option to select skype video call Mar 20 08:50:27 But the default one indicator does that 'ticks' also, so i think it's equal to Mar 20 08:51:10 thedead1440: I can't skype-video anyone, doesn't matter is he able to receive or not Mar 20 08:51:26 Simply no way to initiate call with video Mar 20 08:51:30 XATRIX: shouldn't be so; i've initiated skype video calls several times Mar 20 08:51:42 ask someone to call you via skype video Mar 20 08:51:50 I can only enable video during the call that was made from the outside of Mar 20 08:52:06 then try if selecting that person from recent calls shows skype video Mar 20 08:52:15 thedead1440: 'ask someone to call' :) that way it works :) Mar 20 08:52:20 ah Mar 20 08:52:24 for me it works either way Mar 20 08:52:27 I need to be able to make video-calls by myself Mar 20 08:52:35 usually i initiate it from contacts app myself Mar 20 08:52:45 Alright i'll try now Mar 20 08:56:01 You're right, it's possible to start video call from the contacts Mar 20 08:56:09 :) Mar 20 08:56:12 I tried it from 'last calls' Mar 20 08:56:21 ah that's not accurate then Mar 20 08:56:29 You know, when you open your phone app Mar 20 08:56:33 last calls shows the number that called you and allows you to call back that number Mar 20 08:56:49 Yea Mar 20 08:56:56 What about the dnd more ? Mar 20 08:56:59 f.e. a contact of mine has 3 numbers and calls me from the first number when i want to call back only the first number is allowed Mar 20 08:57:11 If you disable sound/vibra in Profiles Mar 20 08:57:19 It will disable sound completely Mar 20 08:57:27 yup Mar 20 08:57:58 I need to disable sound/vibra/notifications for IM while i'm in 'Red' dnd mode Mar 20 08:58:11 All the other sounds (non-IM) should work as well Mar 20 08:58:43 hmm not sure of that but just choose general profile then with no vibra and silent the im volume Mar 20 08:59:01 DocScrutinizer05: If i set the output power of wifi from 100mW -> 10mW will it help to save charge ? Mar 20 08:59:31 thedead1440: Not a best way, if use go Silent, you're completely out of notifications Mar 20 09:00:25 you don't want notifications for IM right? So in general silencing IM allows you to have all other notifications except for IM right? Mar 20 09:01:00 Yeap, it's should be as you discribed Mar 20 09:01:18 Currently if i'm in DND mode for IM, i'm still have all the notifications enabled Mar 20 09:01:33 yeah so go to settings->profiles-> and in the general tab lower down the volume for IM Mar 20 09:01:35 But there's simply no chat opens while you receive the messages Mar 20 09:02:10 AD-N770: better than wi fi i think dimm lcd lightning... Mar 20 09:02:13 DND mode doesn't affect your system profile. Think of the availability as a status to show not for your phone to act on. Mar 20 09:02:28 Nope, if you decrease the volume for IM in settings, you simply disable the sound , and for BOTH modes, Online+DND Mar 20 09:02:46 ah that's what you mean Mar 20 09:02:54 Yeap Mar 20 09:03:05 It should act like the desktop does Mar 20 09:03:15 i think you would have to use dbus to get the status and then change the volumes. Not sure but check Phone Control Mar 20 09:03:21 http://wiki.maemo.org/Phone_control Mar 20 09:03:42 If you're in DND - no popup notifications, no sound notif. no visual. But the main sounds works, as well as the sound of other apps Mar 20 09:04:34 no idea how to make that work. Like I said DND on device is simply meant to show others your status but not something for your phone to act on Mar 20 09:04:48 You gotta be kidding me, :) i'm not dev. It's like a SDK for dev guys :) Mar 20 09:05:12 Well if you want to customize things you should be willing to learn, shouldn't you? "_" Mar 20 09:06:09 :) Mar 20 09:06:37 Yea, but if i want to customize things, i don't have to actually become a coder :) Mar 20 09:06:57 * XATRIX starts to recall in memory the years of C++/asm Mar 20 09:11:00 AD-N770: sorry tabfail Mar 20 09:11:20 From what i understand of things IM status was not meant to affect your phone. This is carried on in the N9 too. I can have multiple IM accounts and change each of their status' separately. So if you want to do something that wasn't an intention then you'll certainly have to work-around it :) AFAIK dbus signals are the way to go sadly. On the n9 i gave up on dbus and used mc-tool when i wanted to read the current status of specific accounts; Mar 20 09:11:21 maybe something similar on the n900 could help Mar 20 09:11:53 Yea Mar 20 09:11:59 Thanks anyway Mar 20 09:16:03 XATRIX: Wifi power 10/100 doesn't make much of a difference in real life Mar 20 09:16:49 I think even 10 percent of power should drastically drain cell Mar 20 09:17:04 huh? Mar 20 09:18:12 Every +10% i think affects greatly on charge Mar 20 09:18:18 no, WIFI isn't using much energy from battery, since it's not sending all the time, and even if it would, 100mW are not really that much Mar 20 09:20:07 Are you sure ? Even stationary WIFI access-points, use 100mW Mar 20 09:20:18 But it has a PSU ~1-2A 5V Mar 20 09:22:47 that's for the CPU and other things Mar 20 09:24:49 And N900 use no CPU ? :) Mar 20 09:25:36 Or you mean it should also power up FEthernet port, and the whole base board itself inside the access-point ? Mar 20 09:33:35 Where the files for localization /etc/systemui/systemiu.xml ? Mar 20 09:33:44 "powerup_memainmenu_poweroff" and etc... Mar 20 09:39:34 http://tablets-dev.nokia.com/ is dead for me. Where else I can get firmware images? Mar 20 09:39:58 ~skeiron Mar 20 09:39:58 i guess skeiron is the semi-official backup and emergency standin for all internet borne maemo resources: http://skeiron.org/tablets-dev/ http://talk.maemo.org/showthread.php?p=1315143#post1315143 Mar 20 09:40:04 rooster_rus: ^^^ Mar 20 09:43:02 thanks a lot Mar 20 09:43:22 np Mar 20 09:45:46 mmc1: new high speed MMC card at address 0001 - it's a simple SD card inside the handset ? Or it's a soldered die on MB ? Mar 20 09:46:20 when the video player refuses a file... is there a way to get a more exact error message? Mar 20 09:46:33 like, "This codec is not supported" or "This parameter violates H.264 Level 3.0"? Mar 20 09:46:59 one way might be using gst-launch to get a similar playback chain running... has anyone done that yet? Mar 20 09:47:16 (I don't know how to access the HW decoder via gst-launch) Mar 20 09:47:58 10:16:03 DocScrutinizer05 | XATRIX: Wifi power 10/100 doesn't make much of a difference in real life Mar 20 09:48:01 makes virtually none at all Mar 20 09:48:05 because it just sets the max power Mar 20 09:48:19 and wifi is already designed to always use the "minimum possible without too much loss" Mar 20 09:48:40 so by limiting to 10, either nothing changes, OR you will get more packet loss and/or leave the network Mar 20 09:48:51 only in the latter case you'd actually save power :P Mar 20 09:48:54 divVerent: never heard about sucha a "design" Mar 20 09:49:02 the wifi power management always does it Mar 20 09:49:09 transmit power is dynamically adjusted while you are inside Mar 20 09:49:15 the farther you are from the AP, the higher is the power Mar 20 09:49:21 Hmm Mar 20 09:49:37 it dynamically adjusst based on measurements e.g. of signal/noise ratio of the AP's signal Mar 20 09:50:13 so to get most battery drain, run a download with the N900 in a partial faraday cage, blocking as much of the signal as you can without interrupting the download ;) Mar 20 09:50:46 on normal linux systems you actually can watch the power adjustment in iwconfig, have not checked that on the N900 yet Mar 20 09:51:16 hmm Mar 20 09:51:19 also, the N900 has generally quite good wifi reception compared to the iPhone 4, at least it shows a LOT more networks when scanning ;) Mar 20 09:51:24 maybe you have a point Mar 20 09:52:03 Acer Aspire One d255e Mar 20 09:52:07 wlan0 Power Management:off Mar 20 09:52:17 Link Quality=70/70 Signal level=-38 dBm Mar 20 09:52:34 and Mar 20 09:52:35 Bit Rate=65 Mb/s Tx-Power=19 dBm Mar 20 09:53:20 in case it does NOT auto adjust, you may be able to tell the chipset to do it by iwconfig wlan0 txpower auto Mar 20 09:53:32 but if it's 19 for you, it probably IS audo adjusted Mar 20 09:53:49 I currently have 20 dBm transmit power and 41/100 link quality, it clearly is at the max power level Mar 20 09:55:22 Hm Mar 20 09:55:54 http://fpaste.org/hH5j/ - that's for my N900 Mar 20 09:56:08 10dBm is 10mW Mar 20 09:56:13 20dBm is 100mW Mar 20 09:56:23 so you currently are limited to 10mW and the signal is good Mar 20 09:56:40 And actually i've powered it down maually to 10mW Mar 20 09:56:44 So, switching to 100 Mar 20 09:57:07 it also may take a while to adjust, and I am not sure if you can always see the txpower difference Mar 20 09:58:55 Bit Rate=54 Mb/s Tx-Power=20 dBm Mar 20 09:59:00 Power Management:on Mar 20 09:59:05 Link Quality=70/100 Signal level:-49 dBm Noise level=-86 dBm Mar 20 09:59:06 I am currently trying to see if txpower actually changes Mar 20 09:59:12 or if one can't see what the card internally does Mar 20 09:59:25 I know the chips DO the txpower management, but I am trying to see how to find out Mar 20 10:00:17 Passed a minute still 20dBm Mar 20 10:01:07 yes, same here, wondering if one can even ask the actual power Mar 20 10:01:18 That's the difference bettern 100mW (http://fpaste.org/UaoM/) and 10mW (http://fpaste.org/hH5j/) Mar 20 10:01:25 from "iw": Mar 20 10:01:27 dev set txpower [] Mar 20 10:01:29 Specify transmit power level and setting type. Mar 20 10:01:30 DocScrutinizer05: maybe you'll pay some interest as well Mar 20 10:02:07 basically, similar quality level... I would guess also similar actual txpower :P Mar 20 10:02:09 Yep, but i suppose it to adjust automatically Mar 20 10:02:36 "When i connect to the manager with username/passwod and i try to regenerate the setup contact and i see the the wheel that round but i never see the green writing of success. I also can't see in the console the indicator." Mar 20 10:02:40 Gotta love some bug reports Mar 20 10:02:45 Yeap, but with 10mW it was the same quality level but 48Mbit\s instead of 54 Mar 20 10:03:00 But 10mW is less than 100 in 10 times!!! Mar 20 10:03:01 probably changes to 54 too Mar 20 10:03:03 sometimes Mar 20 10:03:07 So, 10 times less drain Mar 20 10:03:11 I am saying, it doesn't ACTUALLY use the 100mW Mar 20 10:03:16 I just want to find a way to prove it :P Mar 20 10:03:48 But it constantly gives me 20dBM Mar 20 10:03:53 And doesn't go down Mar 20 10:04:28 I am trying to find out if the iwconfig value actually reads the status Mar 20 10:04:34 of if it just shows the set limit Mar 20 10:05:09 I think it does the actual Mar 20 10:05:16 then why would it never change Mar 20 10:05:38 even when I explicitly tell it to do auto on my notebook via "iw" Mar 20 10:05:44 Because it does adjust it onece at start Mar 20 10:06:04 I think the default value is auto Mar 20 10:06:25 Bit Rate=48 Mb/s Tx-Power=20 dBm Mar 20 10:06:28 I changed nothing Mar 20 10:06:34 Link Quality=73/100 Signal level:-47 dBm Noise level=-92 dBm Mar 20 10:06:47 the "iw" tool distinguishes between setting the upper limit, or forcing the value Mar 20 10:07:14 iw dev wlp2s0 set txpower limit 1400 sets a limit of 14dBm Mar 20 10:07:23 but "fixed 1400" forces the value Mar 20 10:07:45 but it never changes when read back for me Mar 20 10:07:56 Hm Mar 20 10:09:02 Altering tx power on the n900 does f* all in terms of power saving. Mar 20 10:09:12 vi__: yes, that is the experimental result :P Mar 20 10:09:25 90mW isn't all that much anyway... even if it DOES anything, it's like 30mA Mar 20 10:09:25 At worst it just ruins the SNR Mar 20 10:09:50 but I would still like to see if WiFi auto txpower works on the N900's chip or not Mar 20 10:10:03 I don't think so Mar 20 10:10:04 The thing that allows for most savings is sleep timeout etc. Mar 20 10:10:27 I i still believe that 10mW drains 10 times less energy than 100mW Mar 20 10:10:37 Because it hardware limited Mar 20 10:10:38 only in worst case Mar 20 10:10:42 you're not even sending permanently Mar 20 10:10:52 I'm in idle mode currently Mar 20 10:10:52 yes but 100mW does not give you 10 times the range. Mar 20 10:11:08 Only 5IM protocols in action ,and handset is in standby mode Mar 20 10:11:32 And the most power savings happen when the the TX is off. Mar 20 10:12:01 vi__: no, i'm not talking about the range, for the 'ranged' environments i use GPRS data connection Mar 20 10:12:14 I'm after prolonging the cell life Mar 20 10:12:22 ? Mar 20 10:12:27 battery cell? Mar 20 10:12:32 Actual time to discharge Mar 20 10:12:34 Yea Mar 20 10:12:46 With wifi on? Mar 20 10:13:03 actually, gsm radio is the power hog. Mar 20 10:13:10 Wifi is just fine. Mar 20 10:13:14 I mean, how long i can stay without recharging, and yes, wifi is on, GPRS is also auto-connectable Mar 20 10:13:45 My handset search every 10 minutes, and auto-connect to any available GPRS/WIFI Mar 20 10:13:50 That depends entirely on how much you ise it! Mar 20 10:14:07 Only for IM Mar 20 10:14:13 No web-surfing Mar 20 10:14:17 Wow, that is going to be brutal on power. Mar 20 10:14:39 I just need to be constantly online via IM, nothing extra Mar 20 10:15:01 I think IM doesn't deal that much ? Sends only keepalives while in idle Mar 20 10:16:25 Well, I am connected via wifi all day and I get around 1.5-2 days depending on useage. Mar 20 10:17:11 iwconfig wlan0 rate 12M txpower 20 power period 1 unicast power timeout 20u Mar 20 10:17:25 ^is what I use when streaming internet radio. Mar 20 10:18:03 It takes power usage from ~250mA down to ~90mA Mar 20 10:18:08 vi__: Do you use some IM protocols or any network software ? Mar 20 10:18:27 I think this is due to internet radio somehow turning PSM off for the wifi. Mar 20 10:18:58 I have google talk, skype and a SIP account logged in all the time. Mar 20 10:19:49 Are you on GRPS/3G or only Wifi ? Mar 20 10:20:45 wifi 90% of the time. Mar 20 10:21:13 Hmm Mar 20 10:21:27 Don't know even what to say Mar 20 10:21:38 what is the problem? Mar 20 10:22:19 CSOC: 97 % RSOC: 97 % Mar 20 10:22:20 Average Current: 186 mA Mar 20 10:22:26 But it's currently chargin Mar 20 10:22:56 USB charge Mar 20 10:23:50 also , what's the command to enable PowerManagement on ? Mar 20 10:24:19 From shell... Mar 20 10:24:27 iwconfig wlan0 power on Mar 20 10:24:29 I think. Mar 20 10:25:02 I'm trying to enable it on my laptop Mar 20 10:25:06 Does not seem to throw an error. Mar 20 10:25:14 Just to experiment with Mar 20 10:25:17 root@HP-Compaq-nc4010:~# iwconfig wlan0 power on Mar 20 10:25:17 Error for wireless request "Set Power Management" (8B2C) : Mar 20 10:25:17 SET failed on device wlan0 ; Operation not supported. Mar 20 10:25:34 is the wifi device on? up? in use? Mar 20 10:25:46 Sure, i'm currently via wifi Mar 20 10:25:55 Sounds if my chip doesn't support PM Mar 20 10:26:04 Broadcom Corporation BCM4313 802.11b/g/n Wireless LAN Controller Mar 20 10:26:10 Depends how new the chip is. Mar 20 10:26:27 ok, one thing is for sure: iwconfig can NOT possibly read the current txpower Mar 20 10:26:40 O_o Mar 20 10:26:42 somewhat obvious from the cfg80211_wext_giwtxpower code Mar 20 10:26:56 divVerent: any idea how to get tx ? Mar 20 10:27:02 I am checkign kernel source for that Mar 20 10:27:05 what do you mean? Mar 20 10:27:16 wlan0 IEEE 802.11bg ESSID:"edinst" Mar 20 10:27:16 Mode:Managed Frequency:2.422 GHz Access Point: 68:7F:74:01:D6:A2 Mar 20 10:27:16 Bit Rate=18 Mb/s Tx-Power=20 dBm Mar 20 10:27:16 Retry min limit:7 RTS thr:off Fragment thr=2352 B Mar 20 10:27:27 the 20dBm is the limit, not the current power Mar 20 10:27:44 why do you care about the current power? Mar 20 10:28:10 why not? Mar 20 10:28:50 Yea, i'm also care about current power Mar 20 10:29:19 Well, there is a murata test connector on the PCB... Mar 20 10:29:33 if I am right, the wLAN driver is wl12xx Mar 20 10:29:38 y Mar 20 10:30:11 vi__: Are you kidding :) I'd like not to disassemble the handset and apply multimeter :) Mar 20 10:31:28 Then you clearly do not want to know the TX power SO much. Mar 20 10:31:41 I do :) Mar 20 10:32:02 Cause i want to optimize my power drain Mar 20 10:32:17 one OTHER way BTW would be setting up an adhoc WLAN Mar 20 10:32:22 I think you are barking up the wrong tree. Mar 20 10:32:24 then seeing on the OTHER host the receive power Mar 20 10:32:35 by that you would see if it even changes if you change the txpower limit between 10 and 100 Mar 20 10:32:44 The real power hoover is the screen, then the CPU. Mar 20 10:32:51 sure, I know that Mar 20 10:32:54 for me it's just interesting Mar 20 10:33:15 also, I do like limiting tx power because nobody knows how harmful it is or not Mar 20 10:33:21