Comment 1 for bug 1435109

From IRC:
<rsalveti> Mar 22 23:06:34 ubuntu-phablet powerd[964]: void {anonymous}::reset_alarm_clock() found 3 remaining wakeup requests
<rsalveti> Mar 22 23:06:34 ubuntu-phablet powerd[964]: setting hardware wakeup time to 2015-03-22 23:09:20 for ubuntu push client
<Chipaca> rsalveti: i'm afraid i don't understand what i'm supposed to be gleaning from the kernel logs :(
<Chipaca> rsalveti: i can tell you what the push client should be doing
<rsalveti> so ubuntu push client requested a hardware wakeup time to 2015-03-22 23:09:20, phone suspended fine, woke up at the exact time, then push client got a SysState lock and only released at 23:12:10
<Chipaca> rsalveti: and i _think_ it's doing what it should be doing :)
<Chipaca> rsalveti: push requests a wake up for 5m in the future
<Chipaca> rsalveti: when it wakes up, it asks for a wake lock
* loicm has quit (Ping timeout)
<Chipaca> rsalveti: it then waits up to 1 minute for the network to come up
<Chipaca> rsalveti: when the network comes up, it waits up to 3 minutes for polld to poll stuff
<Chipaca> rsalveti: once polld signals it's done, it asks for a wakeup in 5 minutes, and releases the wake lock
<rsalveti> right, then it is indeed the expected behavior
<Chipaca> phew :)
<rsalveti> but it's annoying that it blocks the suspend for so long
<Chipaca> rsalveti: all those timeouts are configurable
<rsalveti> can't we optimize this?
<Chipaca> rsalveti: you mean, changing those timeouts?
<Chipaca> rsalveti: sure
<rsalveti> Chipaca: when you wait for the network, do you wait the full 60 seconds or do you wait until you get the connection, if less than 60 seconds?
<Chipaca> rsalveti: the latter
<pedronis> the big chunk is polld I suppose
<Chipaca> pedronis: or maybe the network coming up
<rsalveti> network should return quite fast actually
<Chipaca> or all of the above
<rsalveti> 1 minute is a lot of time
<Chipaca> rsalveti: yes
<rsalveti> Chipaca: maybe polld is taking more time now because of the amount of users we have?
<Chipaca> rsalveti: so it's entirely possible there are bugs or unexpected interactions going on; debug logs for push, and the account-polld log, might shed more light.
<Chipaca> rsalveti: polld does not go via our servers
<Chipaca> it's straight to twitter/google
<rsalveti> enabled debug in push-client, how to enable debug for polld?
<Chipaca> rsalveti: do you have a google or twitter account enabled on the device you're checking?
<rsalveti> Chipaca: I got telegram, gmail and twitter
<rsalveti> and I joined the warthogs group in telegram
<Chipaca> rsalveti: gmail and twitter are polld; telegram uses push