Metrics

	A	B	C	D	E	F	G	I
1		Metric	Type	Label	Label Values	Supported on Windows/Linux	Notes (based on current Collector implementation)
2
3	CPU	host/cpu/usage* (seconds)	COUNTER	state	idle	Both
4	*also includes				user	Both
5	cpu label				system	Both
6					interrupt	Both	a.k.a. irq
7					nice	Linux
8					softirq	Linux
9					steal	Linux
10					wait	Linux	a.k.a. iowait (note this could be useful on Windows - see if we can get it somehow)
11		host/cpu/utilization* (%)	GAUGE	(as above)	(as above)	(as above)
12
13	Memory	host/memory/usage (bytes)	GAUGE	state	used	Both
14					free	Both
15					buffered	Linux
16					cached	Linux	(could be obtained on Windows via perf counters, can get free memory ==> Available = Total - Used, Cached = Available - Free)
17					slab_reclaimable	Linux
18					slab_unreclaimable	Linux
19		host/memory/percent_used (%)	GAUGE	(as above)	(as above)	(as above)
20
21	Swap	host/swap/usage* (pages)	GAUGE	state	used	Both	Note this is pagefile usage on Windows
22		host/swap/usage* (pages)	GAUGE	state	free	Both	Note this is pagefile usage on Windows
23	*also includes	host/swap/utilization* (%)	GAUGE	(as above)	(as above)	(as above)
24	device label	host/swap/page_faults (1)	COUNTER	type	major	Linux	On windows, can get total page faults via perf counter, but this may be misleading as this includes minor & major faults, and there doesn't seem to be any way to distinguish these
25		host/swap/page_faults (1)	COUNTER	type	minor	Linux
26		host/swap/paging (1)	COUNTER	type	major	Both	(uses perf counter on Windows)
27				type	minor	Linux
28				direction	in	Both
29				direction	out	Both
30
31	Disk	host/disk/usage* (bytes)	COUNTER	direction	read	Both
32		host/disk/usage* (bytes)	COUNTER	direction	write	Both
33	*also includes	host/disk/ops* (1)	COUNTER	direction	read	Both
34	device label	host/disk/ops* (1)	COUNTER	direction	write	Both
35		host/disk/time* (ms)	COUNTER	direction	read	Both	TODO: confirm why currently listed as GAUGE
36		host/disk/time* (ms)	COUNTER	direction	write	Both
37		host/disk/merged* (1)	COUNTER	direction	read	Linux
38		host/disk/merged* (1)	COUNTER	direction	write	Linux
39
40	Filesystems	host/filesystem/usage* (bytes)	GAUGE	state	used	Both
41					free	Both
42	*also includes				reserved	Linux
43	device label	host/filesystem/utilization* (%)	GAUGE	(as above)	(as above)	(as above)
44		host/filesystem/inodes/used* (1)	GAUGE	state	used	Linux
45		host/filesystem/inodes/used* (1)	GAUGE	state	free	Linux
46		host/filesystem/inodes/utilization* (%)	GAUGE	(as above)	(as above)	(as above)
47
48	Network	host/network/dropped_packets (1)	COUNTER	direction	transmit	Both
49		host/network/dropped_packets (1)	COUNTER	direction	receive	Both		TODO: should we also include network interface as a label here?
50		host/network/packets (1)	COUNTER	direction	transmit	Both
51		host/network/packets (1)	COUNTER	direction	receive	Both
52		host/network/errors (1)	COUNTER	direction	transmit	Both
53		host/network/errors (1)	COUNTER	direction	receive	Both
54		host/network/bytes (bytes)	COUNTER	direction	transmit	Both
55		host/network/bytes (bytes)	COUNTER	direction	receive	Both
56		host/network/tcp_connections (1)	GAUGE	state	... as returned by syscall	Both
57
58	Load	host/load/1m	GAUGE	-	-	Both	(samples perf counter on Windows)
59		host/load/5m	GAUGE	-	-	Both	(samples perf counter on Windows)
60		host/load/15m	GAUGE	-	-	Both	(samples perf counter on Windows)
61
62	Per Process	process/cpu/usage (s)	COUNTER	(as per host/cpu/usage)
63		process/memory/usage (bytes)	GAUGE	(as per host/memory/usage)
64		process/disk/usage (bytes)	COUNTER	(as per host/disk/usage)
65		TODO: add other process metrics?
66
67	TODO	Process Count? (by status)
68		Fork and/or Thread Count?					(note there is a Windows perf counter for thread count)
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100