1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
|
powernapd States
* asleep
* powernapd has successfully taken its action, and has not been awoken yet
* system could be in:
* custom (cpu is online, monitors can run)
* powersaving (cpu is online, monitors can run)
* suspended (cpu is offline)
* hibernated (cpu is offline)
* powered off (cpu is offline)
* awake
* cpu is online monitors can run
Monitors
* Network
* WOLMonitor
* event: bind to ports 7 and 9 and watch for magic packet
* response: record timestamp on any WoL packet arrives
wake immediately if asleep
* default: watch udp:7 or udp:9
* UDPMonitor
* event: bind to specified ports and watch for activity
* response: record timestamp on any packet arrives on any of a list of specified udp ports
wake immediately if asleep
* default: no monitored ports
* TCPMonitor
* poll: run every INTERVAL seconds, looking for connections on specified ports
* response: record timestamp on any established connection against a list of specified tcp ports
wake immediately if asleep
* default: no monitored ports (document that monitoring 22|80|443 might be useful)
* User Input
* USBMonitor
* response: record timestamp on any activity on USB input
wake immediately if asleep
* default: keyboard and mouse
* PS2Monitor
* response: record timestamp on any activity on PS2 input
wake immediately if asleep
* default: keyboard and mouse
* System
* Process
* poll: run every INTERVAL seconds, looking for matching processes
* response: record timestamp on any process matching a list of regexes is running
wake immediately if asleep
* default: no monitored processes
* ProcessIO
* response: record timestamp on any process matching a list of regexes having active io
wake immediately if asleep
* default: no monitored processes
* Load
* poll: run every INTERVAL seconds, looking at /proc/loadavg
* response: record timestamp if system load is too high
wake immediately if asleep
* default: system_load / num_cpus > .75
Approach 1:
* Run all monitors all the time
* Check all monitor timestamps every INTERVAL seconds
* Take powernap action (powersave by default) if no timestamp in last ABSENT_SECONDS
* Continue running monitors in powernap mode
* Wake from powernap any time any timestamp from any running monitor shows up
* This is potentially very expensive
Approach 2:
* Run awake-monitors while awake, asleep-monitors while asleep
Approach 3:
* Run all monitors at all times.
* During ABSENT_SECONDS, check activity every INTERVAL seconds
* During powersave, do NOT check every INTERVAL seconds, but use an EVENT based approach.
* Example:
- While in ABSENT_SECONDS TCPMonitor will be checked for activity every INTERVAL_SECONDS (10)
- If activity was found within the last INTERVAL_SECONDS, then reset absent seconds of monitor.
- While in POWERSAVE mode, the powernapd will stop polling, and TCPMonitor will send an EVENT
- This EVENT will instruct machine to take RECOVER ACTION.
|