948741 - More logging for Android x86 emulator unit tests

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: CIDuty, task)

Description

[Geoff Brown [:gbrown]]

A couple of refinements to the android_emulator_unittest.py logging, in case it helps us understand intermittent loss of connectivity bugs:
 - add "socket" to the list of emulator debug tags
 - when checking emulator health, telnet to the emulator and issue some status commands

Comment 1

[Geoff Brown [:gbrown]]

Attachment: add "socket" to emulator debug tags

Comment 2

[Geoff Brown [:gbrown]]

Attachment: telnet to emulator and check status

Comment 3

[Geoff Brown [:gbrown]]

The first patch adds this type of emulator output:

16:45:08     INFO - control_global_accept: just in (fd=44)
16:45:08     INFO - control_global_accept: creating new client
16:45:08     INFO - control_global_accept: new client 0x2f28c70
16:45:08     INFO - in control_client read: received avd status
16:45:08     INFO - 
16:45:08     INFO - in control_client read: received redir list
16:45:08     INFO - 
16:45:08     INFO - in control_client read: received network status
16:45:08     INFO - 
16:45:08     INFO - in control_client read: received quit

Comment 4

[Geoff Brown [:gbrown]]

The second patch provides output like this:

17:06:28     INFO -   Attempt #1 to connect to emulator on port 5554
17:06:28     INFO - Connected to port 5554
17:06:28     INFO - Android Console: type 'help' for a list of commands
17:06:28     INFO - OK
17:06:28     INFO - avd status: 
17:06:28     INFO - virtual device is running
17:06:28     INFO - OK
17:06:28     INFO - redir list: 
17:06:28     INFO - tcp:20701 => 20701
17:06:28     INFO - tcp:20700 => 20700
17:06:28     INFO - OK
17:06:28     INFO - network status: 
17:06:28     INFO - Current network status:
17:06:28     INFO -   download speed:          0 bits/s (0.0 KB/s)
17:06:28     INFO -   upload speed:            0 bits/s (0.0 KB/s)
17:06:28     INFO -   minimum latency:  0 ms
17:06:28     INFO -   maximum latency:  0 ms
17:06:28     INFO - OK

Comment 5

[Armen [:armenzg]]

Comment on attachment 8345618 [details] [diff] [review]
telnet to emulator and check status

Thank you for keep hammering at this. Best of luck!

Comment 6

[Geoff Brown [:gbrown]]

https://hg.mozilla.org/build/mozharness/rev/6d8d33919a2e
https://hg.mozilla.org/build/mozharness/rev/a47367692ea6

Comment 7

[[e-mail] (:bhearsum)]

(In reply to Geoff Brown [:gbrown] from comment #6)
> https://hg.mozilla.org/build/mozharness/rev/6d8d33919a2e
> https://hg.mozilla.org/build/mozharness/rev/a47367692ea6

I see that these were landed as NPOTB, but is that actually true? I see them referenced in our test configs.

Comment 8

[Armen [:armenzg]]

(In reply to Ben Hearsum [:bhearsum] from comment #7)
> (In reply to Geoff Brown [:gbrown] from comment #6)
> > https://hg.mozilla.org/build/mozharness/rev/6d8d33919a2e
> > https://hg.mozilla.org/build/mozharness/rev/a47367692ea6
> 
> I see that these were landed as NPOTB, but is that actually true? I see them
> referenced in our test configs.

They're only enabled on Cedar. The NPOTB is to remove any worries that this could affect any tier-1 jobs running on other branches.

Comment 9

[[e-mail] (:bhearsum)]

In production

Comment 10

[Armen [:armenzg]]

gbrown: we've decided not to use NPOTB as part of the commit message if the jobs are running anywhere on tbpl. I started using "NPOTB" in this manner so my apologies for doing so.

Comment 11

[Geoff Brown [:gbrown]]

Sorry for the NPOTB confusion here. I won't use it again.

This is working fine on cedar now.