#qi-hardware IRC log for Saturday, 2017-02-04

DocScrutinizer05the whole thing looks like a big fake. Look, 3D-printed ferrite toroid (maybe still plausible) but then a 2N3055 powered from **a 9V-block battery** to drive two 'LAMP' which are at least LED 1W each00:05
DocScrutinizer05prolly the electronic circuit in the LED lamps just works with 9V and the whole circuit is a big fake ;-P00:06
DocScrutinizer05see http://laserhacker.com/?p=52700:07
DocScrutinizer05then WHAT THE FUCK?! he places a coaxial coil (another torus) into the hollow ferrite torus??? 00:10
DocScrutinizer05https://youtu.be/M9Hn5sfMH1Q?t=127  BULLSHIIIIIT00:11
DocScrutinizer05ok, you _can_ do this, but...00:14
DocScrutinizer05HAHAHAHA, I just thought what gonna happen when they reversed the battery, so C becomes E and E becomes C. Circuit looks way better then ;-P00:18
DocScrutinizer05but still will fry the transistor from way too high reverse U_be00:19
DocScrutinizer05no friggin idea what's going on, I guess the circuit diagram is plain wrong00:20
kyakdamnit.. top and iotop show nothing, and still i get 0.5 LA.01:55
kyaki know a reboot would help, but this time i want to figure out what's going on01:56
kyakkilling all heavy-weight suspects, like postgresql, still nothing01:57
kyaki noticed it is related to kernel upgrades not folowed by a reboot01:59
kyakon Arch Linux, kernel files are upgraded in-place.. although i don't understand how this can be aproblem for already loaded modules01:59
DocScrutinizer05zticky topic02:05
DocScrutinizer05tr*02:05
wpwrakkyak: embrace the reboot. don't fight it, my son.02:05
kyak--)02:06
kyaki'll be stripping down this system from running processes till it dies02:07
kyakwhen nothing is left, then it's the kernel02:08
kyakLA strangely starts at 00:00, but the temperature readings don't correlate02:09
kyakcan kernel report LA incorrectly?02:09
kyakhttp://imgur.com/a/h5ads02:10
kyakignore the high load in the middle of the day, i was doing things02:10
kyakthese periodic peaks are also fine, i know what this is02:11
wpwrakload average can/could have a high constant offset if there are processes that want to run but are stuck somewhere02:11
wpwraki had that happen with NFS. not sure if the calculation is still the same, though02:12
kyakhow these processes will exhibit themselves in top?02:12
kyaki don't nfs, but have smb mounted.. let's try unmounting that02:12
wpwrakthere may be unusual flags in STAT of ps aux02:13
wpwrak(unmount) if you can ;-)02:14
kyakyou nailed it! it was a stall smb mount (remote host is not available)02:15
wpwrakheh :)02:15
kyakyeah, that's it.. Thanks!02:15
DocScrutinizer05LA?02:48
DocScrutinizer05err yes, I see lsysguard report bogus crap for basically all values based on /proc/* (the latter my suspicion), after a while since start of the monitor. I rend to blame systemd and the crap with cgroups and mounts it does02:50
DocScrutinizer05ksysguard that is02:50
DocScrutinizer05and a 2nd yes: processes idling in kernel IO can add to load while not using any CPU cycles. No idea where to spot them, iotop maybe02:54
DocScrutinizer05or "wchan" in top02:54
kyakyeah, it's hard to spot those.. Especially because this is not a process, but a kernel thread, like in my case of samba mount02:56
kyakit just didn't reveal itself in iotop02:56
DocScrutinizer05(kernel thread) haha, particularly nasty, yes02:58
DocScrutinizer05yeah, prolly kswapd etc would be able to create similar shite02:59
DocScrutinizer05http://paste.opensuse.org/61614886  pstree 203:03
DocScrutinizer05filesystems via USB are particularly nasty regarding this, can freeze complete system03:04
DocScrutinizer05F U systemd, wchan does only show "-" in htop03:11
DocScrutinizer05nfc, maybe my htop is bitrotten03:13
DocScrutinizer05ps -l works03:13
DocScrutinizer05top also works03:15
DocScrutinizer05might be the cold I caught03:16
DocScrutinizer05"Memorys Vulnerability" - excellent!  http://abcnews.go.com/Technology/story?id=98195&page=104:00
--- Sun Feb 5 201700:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!