[mako] Awful life battery since OTA 13

I also add syslog and repowerd from /var/log

and syslog

Could you please also attach the result of (i.e. the /tmp/top.txt file):

top -b -n10 -d1 > /tmp/top.txt

on the device that exhibits this behavior.

An interesting snippet from the syslog:

// Suspend
Oct 4 18:14:26 ubuntu-phablet kernel: [20994.425695] suspend: enter suspend
Oct 4 18:14:26 ubuntu-phablet kernel: [20994.425756] PM: suspend entry 2016-10-04 16:14:26.176861138 UTC
Oct 4 18:14:26 ubuntu-phablet kernel: [20994.425878] PM: Syncing filesystems...
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.507337] Freezing user space processes ... (elapsed 0.001 seconds) done.
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.526535] sync done.
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.749455] Freezing remaining freezable tasks ... (elapsed 0.003 seconds) done.
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.752782] Suspending console(s) (use no_console_suspend to debug)
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.764166] msm_fb_ext_suspend: Turning off HPD circuitry
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.803446] PM: suspend of devices complete after 46.878 msecs
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.805216] PM: late suspend of devices complete after 1.739 msecs
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.808878] active wake lock alarm_rtc, time left 96
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.808909] power_suspend_late return -11
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.808970] dpm_run_callback(): power_suspend_late+0x0/0x68 returns -11
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.809031] PM: Device power.0 failed to suspend noirq: error -11
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.811290] PM: noirq resume of devices complete after 2.227 msecs
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.814952] PM: early suspend of devices complete after 1.739 msecs
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.814983] PM: Some devices failed to power down

// Immediately resume
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.821575] msm_fb_ext_resume: Turning on HPD circuitry
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.827496] PM: resume of devices complete after 12.482 msecs
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.829022] Restarting tasks ... done.
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.869919] PM: suspend exit 2016-10-04 16:14:27.047214477 UTC
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.869950] suspend: exit suspend, ret = -11 (2016-10-04 16:14:27.047214477 UTC)
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.869950] active wake lock alarm_rtc, time left 89
Oct 4 18:14:27 ubuntu-phablet kernel: [20994.869950] suspend: abort suspend
Oct 4 18:14:27 ubuntu-phablet kernel: [20995.761511] wake lock alarm_rtc, expired

... and this is going on continuously while the phone is idle. The phone suspends but it immediately wakes up, seemingly because a very short alarm is set ("active wake lock alarm_rtc, time left XX", XX is in milliseconds), then it suspends again, and wakes up again immediately, and so on. I don't know who is setting this alarm.

https://bugs.launchpad.net/ubuntu-power-consumption/+bug/1603024 could be related.

suspend-blocker confirms, I know of no process that would set the alarm in this way

Suspend blocking wakelocks:
  alarm_rtc 2471 99.92%
  msm_hsic_host 1 0.04%
  power-supply 1 0.04%

Resume wakeup causes:
  None

Suspend failure causes:
  late suspend wakelock 2471 99.96%
  devices failed to suspend 1 0.04%

Time between successful suspends:
   Interval (seconds) Frequency Cumulative Time (Seconds)
  1024.000 - 2047.999 2 66.67% 3745.55 64.37%
  2048.000 - 4095.999 1 33.33% 2072.79 35.63%

Duration of successful suspends:
   Interval (seconds) Frequency Cumulative Time (Seconds)
    32.000 - 63.999 2 50.00% 92.45 12.69%
    64.000 - 127.999 0 0.00% 0.00 0.00%
   128.000 - 255.999 1 25.00% 169.11 23.20%
   256.000 - 511.999 1 25.00% 467.21 64.11%

Suspends:
  2485 suspends aborted (99.84%).
  4 suspends succeeded (0.16%).
  total time: 728.770144 seconds (11.13%).
  minimum: 40.214110 seconds.
  maximum: 467.213566 seconds.
  mean: 182.192536 seconds.
  mode: 467.000000 seconds.
  median: 110.671234 seconds.

Time between successful suspends:
  total time: 5818.346063 seconds (88.87%).
  minimum: 1787.768036 seconds.
  maximum: 2072.791465 seconds.
  mean: 1939.448688 seconds.
  mode: 2073.000000 seconds.
  median: 1957.786563 seconds.

Hi,

Here is the result of the top command
For information, there is no apps launched, and still the battery drains
Thanks

Status changed to 'Confirmed' because the bug affects multiple users.

Hi,
Strange thing, battery does suddenly though it shows something like 1/4 life in the systray
I had syslog and repowerd, I have all other logs if needed

rsyslog and repowerd just before it suddenly dies

I have prepared a PPA containing a new repowerd with a potential fix for the issue here:

https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/2276

IMPORTANT CAVEAT: The changes are experimental. Only try the PPA packages on a development device, not one you care about keeping stable. You may need to reflash a clean image if things go very wrong.

To try the updated packages follow the instructions from the link below:

https://wiki.ubuntu.com/Process/Merges/TestPlans/repowerd

Let me know if/how the fix works for you.

Hi, sorry I've just upgraded to OTA14

Ok i have just tested this silo and i see a great improvement in battery life.

I have noticed previously a great difference in battery consumption when i turn off wifi.
e.g. when turning off wifi consumption is 1% overnight with wifi on it's like 8-12%

with the installed silo i am around 2% which is a great improvment.
It s like 17:07 and i still have 92% battery.

In terms of regression i am not sure but i have the feeling that wifi are not re-connecting when i leave the access point.

@pikapika you can install the app ubuntu silo installer on your phone and install the silo #2276 to test

I have tested

Test that the device can automatically shutdown the screen and suspend
Pass

Test that the device can wake up/suspend when pressing the power button
pass

Test that the device can wake up when receiving a call
pass

Test that the device can wake up when connecting/disconnecting usb charger
pass

Unplug the usb cable and reboot the phone. It shouldn't take more than 30 seconds to reboot to the bootloader.
42sec.

    This is because if the phone takes more than 30 seconds to reboot it might mean that it's trying to suspend at the same time (bug 1452300)

Play a long (>2min) video and:

    Ensure the display stays on after the default blanking timeout has passed (60s)
    *After* step (1) (i.e. after having waited 60s) unplug the USB cable. The display should stay on (bug 1602933) until the video finishes.

Not possible to test oxide bug
https://bugs.launchpad.net/canonical-devices-system-image/+bug/1647895

Test that the phone actually suspends. A good test is to turn off wireless and BT, disconnect from any power sources and turn off the screen. After a while connect to the phone again and check /var/log/syslog. There should be a gap in the syslog entries corresponding to the duration of suspend.
not tested
Was working nice before with wifi off so only observe improvement when wifi is on.

Test that hw alarms are being triggered. Set an alarm (through the alarm clock) for a few minutes in the future and let the device suspend. The alarm should ring at the appropriate time.
pass

This bug was fixed in the package repowerd - 2016.12+17.04.20161212.1-0ubuntu1

---------------
repowerd (2016.12+17.04.20161212.1-0ubuntu1) zesty; urgency=medium

  * New 2016.12 release
    - Support suspending with logind
    - Implement basic lid handling
    - Don't change the display power mode when pressing the power
      button on non-phablet devices.
    - Various snap-related improvements
    - Fix "[mako] Alarm doesn't ring when screen locked" (LP: #1588526)
    - Fix "[mako] Awful life battery since OTA 13" (LP: #1630382)

 -- Alexandros Frantzis <email address hidden> Mon, 12 Dec 2016 19:02:03 +0000