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
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
|
Windows PMDA
============
This PMDA collects performance data from a Microsoft Windows kernel.
Metrics
=======
The help text is exported from the kernel via the PDH (Performance
Data Helper) APIs. To view the help text, install the PMDA, then
the following command will list all the available metrics and their
explanatory "help" text:
$ pminfo -fT kernel disk mem network filesys sqlserver hinv pmda process
Installation
============
+ # cd $PCP_PMDAS_DIR/windows
+ Check that there is no clash in the Performance Metrics Domain
defined in ./domain.h and the other PMDAs currently in use (see
$PCP_PMCDCONF_PATH). If there is, edit ./domain.h to choose another
domain number.
+ Then simply use
# ./Install
and choose both the "collector" and "monitor" installation
configuration options -- everything else is automated.
De-installation
===============
+ Simply use
# cd $PCP_PMDAS_DIR/windows
# ./Remove
Troubleshooting
===============
+ After installing or restarting the agent, the PMCD log file
($PCP_LOG_DIR/pmcd/pmcd.log) and the PMDA log file
($PCP_LOG_DIR/pmcd/windows.log) should be checked for any warnings
or errors.
Adding New Metrics
==================
The following steps should be followed when adding new metrics ... this
assumes the MinGW gcc compiler is being used.
a. Make sure you know what you've currently got, so on the target system
$ ./pdhlist | ./pdhmatch >status-quo
b. Pick the Pdh paths you're interested in (drop the hostname prefix,
add the (*/*#*) or * patterns as appropriate. Beware that metrics
with multiple instances are reported multiple times ... for example
SQLServer:Cache Manager\Cache Hit Ratio
SQLServer:Cache Manager(_Total)\Cache Hit Ratio
SQLServer:Cache Manager(*/*#*)\Cache Hit Ratio
of these, the first pattern must not be used ... use the second one
if you want the "over all instances" totals, and the last one gives
the per instance metrics.
c. Choose a PMID ... use the next unused number in sequence ... check
metricdesc[] in pmda.c, you want the PMDA_PMID(0,x) macro from the
last entry in the table, and the number you're after is x+1 (x+1 =
81 for my example).
d. Choose a name and where to put the new metric in the namespace ...
edit the appropriate pmns.* file (let's assume you're adding to one
of the existing Pdh groups, rather than creating a new one which is
more complicated), adding an appropriate entry using x+1 from above
for the last component of the PMID. Something like this in
pmns.sqlserver
sqlserver {
...
cache_mgr // new line
}
...
// all new lines from here down
sqlserver.cache_mgr {
all
percache
}
sqlserver.cache_mgr.all {
cache_hit_ratio WINDOWS:0:81
}
sqlserver.cache_mgr.cache {
cache_hit_ratio WINDOWS:0:82
}
Write the modified pmns file.
Check you've got it right with
$ pminfo -m -n root sqlserver.cache_mgr
sqlserver.cache_mgr.all.cache_hit_ratio PMID: 79.0.81
sqlserver.cache_mgr.cache.cache_hit_ratio PMID: 79.0.82
e. Metric semantics ... you'll need a complete metricdesc[i]
initializer from pmda.c ... pick one that is similar, copy, paste
at the end of the table initialization and update PMDA_PMID(0,y)
with x+1 from step a.
To help with "similar" here are some examples:
kernel.all.cpu.user - unsigned 64-bit counter in units of microseconds
network.interface.in.bytes - unsigned 32-bit counter in units of
bytes with one value per network interface
mem.available - unsigned 64-bit instantaneous value in units of
Mbytes
sqlserver.buf_mgr.cache_hit_ratio - floating point value
representing a cache hit ratio, returned in the range 0.0 to
1.0
Be careful with '\' in the Pdh patterns. Replace them with '\\' in the
string initializer used here.
My new metrics are cache hit ratios
/* sqlserver.cache_mgr.all.cache_hit_ratio */
{ { PMDA_PMID(0,81), PM_TYPE_FLOAT, PM_INDOM_NULL, PM_SEM_INSTANT,
PMDA_PMUNITS(0, 0, 0, 0, 0, 0) }, M_OPTIONAL, 0, 0, 0, NULL,
"\\SQLServer:Cache Manager(_Total)\\Cache Hit Ratio"
},
/* sqlserver.cache_mgr.cache.cache_hit_ratio */
{ { PMDA_PMID(0,82), PM_TYPE_FLOAT, SQL_CACHE_INDOM, PM_SEM_INSTANT,
PMDA_PMUNITS(0, 0, 0, 0, 0, 0) }, M_OPTIONAL, 0, 0, 0, NULL,
"\\SQLServer:Cache Manager(*/*#*)\\Cache Hit Ratio"
},
f. If you've added a new instance domain (SQL_CACHE_INDOM in the case
above, this needs to be handled):
i. add a new header #define line (use the next ordinal number), e.g.
#define SQL_CACHE_INDOM 5
ii. add a new entry to indomtab[] in pmda.c, e.g.
{ SQL_CACHE_INDOM, 0, NULL },
iii. add a new entry to indomtab[] in instance.c, e.g.
SQL_CACHE_INDOM,
iv. add new code in the switch of check_instance() in instance.c
to parse the Pdh instance name to extract a PCP instance name
and assign an instance number ... it is strongly suggested
that you study the ones already there, steal and modify the
one that is closest to your new instance
g. Make and upgrade (assuming you're on the target machine and not
interested in packaging)
$ make
$ su
# /etc/pcp stop
# make install
# cd $PCP_VAR_DIR/pmdas/windows
# ./Install
|
