1PMSTAT(1) General Commands Manual PMSTAT(1)
2
6 pcp-vmstat, pmstat - high-level system performance overview
7
9 pcp [pcp options] vmstat [interval [samples]]
10 pmstat [-gLlPxz?] [-a archive] [-A align] [-h host] [-H file] [-n
12 pmnsfile] [-O offset] [-p port] [-s samples] [-S starttime] [-t inter‐
13 val] [-T endtime] [-Z timezone]
14
16 pmstat provides a one line summary of system performance every interval
17 unit of time (the default is 5 seconds). pmstat is intended to monitor
18 system performance at the highest level, after which other tools may be
19 used to examine subsystems in which potential performance problems may
20 be observed in greater detail.
21 pcp-vmstat is a simple wrapper for use with the pcp(1) command, provid‐
23 ing a more familiar command line format for some users. It also
24 enables the extended reporting option by default, see the -x option
25 below.
26 Multiple hosts may be monitored by supplying more than one host with
28 multiple -h flags (for live monitoring) or by providing a name of the
29 hostlist file, where each line contain one host name, with -H, or mul‐
30 tiple -a flags (for retrospective monitoring from sets of archives).
31 By default, pmstat fetches metrics by connecting to the Performance
33 Metrics Collector Daemon (PMCD) on the local host. If the -L option is
34 specified, then pmcd(1) is bypassed, and metrics are fetched from PMDAs
35 on the local host using the standalone PM_CONTEXT_LOCAL variant of
36 pmNewContext(3). When the -h option is specified, pmstat connects to
37 the pmcd(1) on host and fetches metrics from there. As mentioned
38 above, multiple hosts may be monitored by supplying multiple -h flags.
39 Alternatively, if the -a option is used, the metrics are retrieved from
41 the Performance Co-Pilot archive log files identified by archive, which
42 is a comma-separated list of names, each of which may be the base name
43 of an archive or the name of a directory containing one or more ar‐
44 chives. Multiple sets of archives may be replayed by supplying multi‐
45 ple -a flags. When the -a flag is used, the -P flag may also be used
46 to pause the output after each interval.
47 Standalone mode can only connect to the local host, using a set of ar‐
49 chives implies a host name, and nominating a host precludes using an
50 archive, so the options -L, -a and -h are mutually exclusive.
51 pmstat may relinquish its own timing control, and operate as a
53 ``slave'' of a pmtime(1) process that uses a GUI dialog to provide tim‐
54 ing control. In this case, either the -g option should be used to
55 start pmstat as the sole slave of a new pmtime(1) instance, or -p
56 should be used to attach pmstat to an existing pmtime(1) instance via
57 the IPC channel identified by the port argument.
58 The -S, -T, -O and -A options may be used to define a time window to
60 restrict the samples retrieved, set an initial origin within the time
61 window, or specify a ``natural'' alignment of the sample times; refer
62 to PCPIntro(1) for a complete description of these options.
63
65 The available command line options are:
66 -a archive, --archive=archive
68 Performance metric values are retrieved from the set of Perfor‐
69 mance Co-Pilot (PCP) archive log files identified by the archive
70 argument, which is a comma-separated list of names, each of which
71 may be the base name of an archive or the name of a directory con‐
72 taining one or more archives.
73 -A align, --align=align
75 Force the initial sample to be aligned on the boundary of a natu‐
76 ral time unit align. Refer to PCPIntro(1) for a complete descrip‐
77 tion of the syntax for align.
78 -g, --guimode
80 Start pmstat as the slave of a new pmtime(1) process for replay of
81 archived performance data using the pmtime(1) graphical user
82 interface.
83 -h host, --host=host
85 Fetch performance metrics from pmcd(1) on host, rather than from
86 the default localhost.
87 -H path, --hostsfile=path
89 Specify the path to a file containing a set of hostnames where
90 pmcd(1) is running , rather than using the default localhost.
91 -K spec, --spec-local=spec
93 When fetching metrics from a local context (see -L), the -K option
94 may be used to control the DSO PMDAs that should be made accessi‐
95 ble. The spec argument conforms to the syntax described in
96 pmSpecLocalPMDA(3). More than one -K option may be used.
97 -l, --suffix
99 Prints the last 7 characters of a hostname in summaries involving
100 more than one host (when more than one -h option has been speci‐
101 fied on the command line).
102 -L, --local-PMDA
104 Use a local context to collect metrics from DSO PMDAs on the local
105 host without PMCD. See also -K.
106 -n pmnsfile, --namespace=pmnsfile
108 Load an alternative Performance Metrics Name Space (PMNS(5)) from
109 the file pmnsfile.
110 -O origin, --origin=origin
112 When reporting archived metrics, start reporting at origin within
113 the time window (see -S and -T). Refer to PCPIntro(1) for a com‐
114 plete description of the syntax for origin.
115 -p port, --guiport=port
117 Attach pmstat to an existing pmtime(1) time control process
118 instance via the IPC channel identified by the port argument.
119 This option is normally only used by other tools, e.g.
120 pmchart(1), when they launch pmstat with synchronized time con‐
121 trol.
122 -s samples, --samples=samples
124 The samples option defines the number of samples to be retrieved
125 and reported. If samples is 0 or -s is not specified, pmstat will
126 sample and report continuously - this is the default behavior.
127 -S starttime, --start=starttime
129 When reporting archived metrics, the report will be restricted to
130 those records logged at or after starttime. Refer to PCPIntro(1)
131 for a complete description of the syntax for starttime.
132 -t interval, --interval=interval
134 Set the reporting interval to something other than the default 1
135 second. The interval argument follows the syntax described in
136 PCPIntro(1), and in the simplest form may be an unsigned integer
137 (the implied units in this case are seconds).
138 -T endtime, --finish=endtime
140 When reporting archived metrics, the report will be restricted to
141 those records logged before or at endtime.
142 -x, --xcpu
144 The extended CPU metrics option, causes two additional CPU metrics
145 to be reported, namely wait for I/O ("wa") and virtualisation
146 steal time ("st").
147 -z, --hostzone
149 Change the reporting timezone to the local timezone at the host
150 that is the source of the performance metrics, as identified via
151 either the -h or -a options.
152 -Z timezone, --timezone=timezone
154 By default, pmtime reports the time of day according to the local
155 timezone on the system where pmstat is run. The -Z option changes
156 the timezone to timezone in the format of the environment variable
157 TZ as described in environ(7).
158
160 The output from pmstat is directed to standard output, and the columns
161 in the report are interpreted as follows:
162 loadavg The 1 minute load average.
164 memory The swpd column indicates average swap space used during the
166 interval, in Kbytes. The free column indicates average free
167 memory during the interval, in Kbytes. The buff column indi‐
168 cates average buffer memory in use during the interval, in
169 Kbytes. The cache column indicates average cached memory in
170 use during the interval, in Kbytes.
171 If the values become large, they are reported as Mbytes (m
173 suffix) or Gbytes (g suffix).
174 swap The metrics in this area of the kernel instrumentation are of
176 varying value. We try to report the average number of pages
177 that are paged in (pi) and out (po) per second during the
178 interval. If the corresponding page swapping metrics are
179 unavailable, we report the average rate per second of swap
180 operations in (si) and out (so) during the interval. It is
181 normal for the ``in'' values to be non-zero, but the system
182 is suffering memory stress if the ``out'' values are non-zero
183 over an extended period.
184 If the values become large, they are reported as thousands of
186 operations per second (K suffix) or millions of operations
187 per second (M suffix).
188 io The bi and bo columns indicate the average rate per second of
190 block input and block output operations (respectfully) during
191 the interval. Unless all file systems have a 1 Kbyte block
192 size, these rates do not directly indicate Kbytes trans‐
193 ferred.
194 If the values become large, they are reported as thousands of
196 operations per second (K suffix) or millions of operations
197 per second (M suffix).
198 system Interrupt rate (in) and context switch rate (cs). Rates are
200 expressed as average operations per second during the inter‐
201 val. Note that the interrupt rate is normally at least HZ
202 (the clock interrupt rate, usually 100) interrupts per sec‐
203 ond.
204 If the values become large, they are reported as thousands of
206 operations per second (K suffix) or millions of operations
207 per second (M suffix).
208 cpu Percentage of CPU time spent executing user and "nice user"
210 code (us), system and interrupt processing code (sy), idle
211 loop (id).
212 If any values for the associated performance metrics are unavailable,
214 the value appears as ``?'' in the output.
215
217 All are generated on standard error and are intended to be self-
218 explanatory.
219
221 $PCP_VAR_DIR/pmns/*
222 default PMNS specification files
223 $PCP_VAR_DIR/config/pmlogger/config.pmstat
224 pmlogger(1) configuration for creating an archive suitable
225 for replay with pmstat
226
228 Environment variables with the prefix PCP_ are used to parameterize the
229 file and directory names used by PCP. On each installation, the file
230 /etc/pcp.conf contains the local values for these variables. The
231 $PCP_CONF variable may be used to specify an alternative configuration
232 file, as described in pcp.conf(5).
233
235 PCPIntro(1), pmclient(1), pmtime(1), PMAPI(3), pmNewContext(3),
236 pcp.conf(5), pcp.env(5) and PMNS(5).
237Performance Co-Pilot PCP PMSTAT(1)