IssueTracker

Yes, that is correct.

Yeah, both logcat and the dump for adb shell dumpsys jobscheduler for your app are going to be very useful.

You need to look at adb logcat.

Can you send me all the contents of adb logcat thats scoped beyond your app ? Once you re-launch the app does the job get scheduled again ? There is unfortunately not a lot of useful information in dumpsys jobscheduler.

Also, which device are you testing this on ?

Looking at your logs, you are getting penalized for not stopping work.

Your Worker continues to run after it had been asked to stop. You need to stop cooperatively or you will get penalized. That's why the OS revoked connectivity for your uid to force you to stop.

My guess is JobScheduler is penalizing your app, and therefore it won't run your work.

Also the output from adb shell dumpsys jobscheduler is not complete. Can you please make sure you dump all the info ?

No, that was always a requirement (since WorkManager 1.0). This behavior is not tied to WorkManager's implementation though. Its JobSchdeduler that drives the work. So something could have changed there.

I don't see the logs for an App launch after the app exited doze mode? Do you have additional logs from that session ?

Thanks. I don't see anything unusual there.

Save this script and try running this using python 3. Pass the fully qualified package name as the argument to the script. So something like: python3 jobs.py <your_package_name>.

I still feel like I am missing huge chunks of dumpsys jobscheduler. Alternatively, you can also do something like: adb shell dumpsys jobscheduler > out.txt and send me that.

import os
import sys


def jobs(name):
    command = 'adb shell dumpsys jobscheduler | grep -A 24 ".*JOB.*%s"' % (name)
    os.system(command)

if __name__ == "__main__":
    if len(sys.argv) < 2:
        print('Usage is ./job.py package_name')
    else:
        name = sys.argv[1]
        jobs(name.strip())

I think you may have forgotten to attach the file.

Nothing looks out of the ordinary. It looks like you had the Worker run a couple of times based on:

Timer{<0>com.simonits.adalerts} NOT active, 0 running bg jobs
  
    Saved sessions:
      996813165 -> 996902031 (88866), 1 bg jobs.
      995442568 -> 995655179 (212611), 1 bg jobs.
      995400126 -> 995442559 (42433), 1 bg jobs.
      991729518 -> 991739652 (10134), 1 bg jobs.
      988324231 -> 988392138 (67907), 1 bg jobs.
      987839386 -> 987894917 (55531), 1 bg jobs.
      973374788 -> 973542710 (167922), 1 bg jobs.
      973335801 -> 973335832 (31), 1 bg jobs.
      973334781 -> 973335795 (1014), 1 bg jobs.
      973298163 -> 973298778 (615), 1 bg jobs.
      973164612 -> 973295300 (130688), 1 bg jobs.
      972009891 -> 972228958 (219067), 1 bg jobs.
      970880825 -> 971109890 (229065), 1 bg jobs.
      969749680 -> 969976527 (226847), 1 bg jobs.
      968608690 -> 968847501 (238811), 1 bg jobs.
      967034243 -> 967091488 (57245), 1 bg jobs.
      965486455 -> 965562052 (75597), 1 bg jobs.
      964497879 -> 964555054 (57175), 1 bg jobs.
      963084431 -> 963303262 (218831), 1 bg jobs.
      961964759 -> 962178948 (214189), 1 bg jobs.
      961282552 -> 961468886 (186334), 1 bg jobs.
      960842943 -> 960902474 (59531), 1 bg jobs.
      959881647 -> 959938712 (57065), 1 bg jobs.
      958821517 -> 958977069 (155552), 1 bg jobs.
      957169812 -> 957383549 (213737), 1 bg jobs.

So your Worker was running.

Your Worker should respect stoppage requests. Try adding that. Apart from that, nothing here looks out of the ordinary. I am marking this bug as not-reproducible.

If you have more information, feel free to comment and I can take a look.

I found an edge case, where your Worker will not correctly be rescheduled if it got interrupted.

That's probably why JobScheduler is not picking it up. Adding a fix.

https://android-review.googlesource.com/c/platform/frameworks/support/+/1246571 should address the problem.

I just wanted to keep you in the loop. This should be addressed by WM 2.3.3.