1PIDSTAT(1) Linux User's Manual PIDSTAT(1)
2
6 pidstat - Report statistics for Linux tasks.
7
9 pidstat [ -d ] [ -h ] [ -I ] [ -l ] [ -r ] [ -s ] [ -t ] [ -U [ user‐
10 name ] ] [ -u ] [ -V ] [ -w ] [ -C comm ] [ -p { pid [,...] | SELF |
11 ALL } ] [ -T { TASK | CHILD | ALL } ] [ interval [ count ] ]
12
14 The pidstat command is used for monitoring individual tasks currently
15 being managed by the Linux kernel. It writes to standard output activ‐
16 ities for every task selected with option -p or for every task managed
17 by the Linux kernel if option -p ALL has been used. Not selecting any
18 tasks is equivalent to specifying -p ALL but only active tasks (tasks
19 with non-zero statistics values) will appear in the report.
20 The pidstat command can also be used for monitoring the child processes
22 of selected tasks. Read about option -T below.
23 The interval parameter specifies the amount of time in seconds between
25 each report. A value of 0 (or no parameters at all) indicates that
26 tasks statistics are to be reported for the time since system startup
27 (boot). The count parameter can be specified in conjunction with the
28 interval parameter if this one is not set to zero. The value of count
29 determines the number of reports generated at interval seconds apart.
30 If the interval parameter is specified without the count parameter, the
31 pidstat command generates reports continuously.
32 You can select information about specific task activities using flags.
34 Not specifying any flags selects only CPU activity.
35
38 -C comm
39 Display only tasks whose command name includes the string comm.
40 This string can be a regular expression.
41 -d Report I/O statistics (kernels 2.6.20 and later only). The fol‐
43 lowing values may be displayed:
44 UID
46 The real user identification number of the task being
47 monitored.
48 USER
50 The name of the real user owning the task being moni‐
51 tored.
52 PID
54 The identification number of the task being monitored.
55 kB_rd/s
57 Number of kilobytes the task has caused to be read from
58 disk per second.
59 kB_wr/s
61 Number of kilobytes the task has caused, or shall cause
62 to be written to disk per second.
63 kB_ccwr/s
65 Number of kilobytes whose writing to disk has been can‐
66 celled by the task. This may occur when the task trun‐
67 cates some dirty pagecache. In this case, some IO which
68 another task has been accounted for will not be happen‐
69 ing.
70 Command
72 The command name of the task.
73 -h Display all activities horizontally on a single line. This is
75 intended to make it easier to be parsed by other programs.
76 -I In an SMP environment, indicate that tasks CPU usage (as dis‐
78 played by option -u ) should be divided by the total number of
79 processors.
80 -l Display the process command name and all its arguments.
82 -p { pid [,...] | SELF | ALL }
84 Select tasks (processes) for which statistics are to be
85 reported. pid is the process identification number. The SELF
86 keyword indicates that statistics are to be reported for the
87 pidstat process itself, whereas the ALL keyword indicates that
88 statistics are to be reported for all the tasks managed by the
89 system.
90 -r Report page faults and memory utilization.
92 When reporting statistics for individual tasks, the following
94 values may be displayed:
95 UID
97 The real user identification number of the task being
98 monitored.
99 USER
101 The name of the real user owning the task being moni‐
102 tored.
103 PID
105 The identification number of the task being monitored.
106 minflt/s
108 Total number of minor faults the task has made per sec‐
109 ond, those which have not required loading a memory page
110 from disk.
111 majflt/s
113 Total number of major faults the task has made per sec‐
114 ond, those which have required loading a memory page from
115 disk.
116 VSZ
118 Virtual Size: The virtual memory usage of entire task in
119 kilobytes.
120 RSS
122 Resident Set Size: The non-swapped physical memory used
123 by the task in kilobytes.
124 %MEM
126 The tasks's currently used share of available physical
127 memory.
128 Command
130 The command name of the task.
131 When reporting global statistics for tasks and all their chil‐
133 dren, the following values may be displayed:
134 UID
136 The real user identification number of the task which is
137 being monitored together with its children.
138 USER
140 The name of the real user owning the task which is being
141 monitored together with its children.
142 PID
144 The identification number of the task which is being mon‐
145 itored together with its children.
146 minflt-nr
148 Total number of minor faults made by the task and all its
149 children, and collected during the interval of time.
150 majflt-nr
152 Total number of major faults made by the task and all its
153 children, and collected during the interval of time.
154 Command
156 The command name of the task which is being monitored
157 together with its children.
158 -s Report stack utilization. The following values may be dis‐
160 played:
161 UID
163 The real user identification number of the task being
164 monitored.
165 USER
167 The name of the real user owning the task being moni‐
168 tored.
169 PID
171 The identification number of the task being monitored.
172 StkSize
174 The amount of memory in kilobytes reserved for the task
175 as stack, but not necessarily used.
176 StkRef
178 The amount of memory in kilobytes used as stack, refer‐
179 enced by the task.
180 Command
182 The command name of the task.
183 -T { TASK | CHILD | ALL }
185 This option specifies what has to be monitored by the pidstat
186 command. The TASK keyword indicates that statistics are to be
187 reported for individual tasks (this is the default option)
188 whereas the CHILD keyword indicates that statistics are to be
189 globally reported for the selected tasks and all their children.
190 The ALL keyword indicates that statistics are to be reported for
191 individual tasks and globally for the selected tasks and their
192 children.
193 Note: Global statistics for tasks and all their children are not
195 available for all options of pidstat. Also these statistics are
196 not necessarily relevant to current time interval: The statis‐
197 tics of a child process are collected only when it finishes or
198 it is killed.
199 -t Also display statistics for threads associated with selected
201 tasks.
202 This option adds the following values to the reports:
204 TGID
206 The identification number of the thread group leader.
207 TID
209 The identification number of the thread being monitored.
210 -U [ username ]
212 Display the real user name of the tasks being monitored instead
213 of the UID. If username is specified, then only tasks belonging
214 to the specified user are displayed.
215 -u Report CPU utilization.
217 When reporting statistics for individual tasks, the following
219 values may be displayed:
220 UID
222 The real user identification number of the task being
223 monitored.
224 USER
226 The name of the real user owning the task being moni‐
227 tored.
228 PID
230 The identification number of the task being monitored.
231 %usr
233 Percentage of CPU used by the task while executing at the
234 user level (application), with or without nice priority.
235 Note that this field does NOT include time spent running
236 a virtual processor.
237 %system
239 Percentage of CPU used by the task while executing at the
240 system level (kernel).
241 %guest
243 Percentage of CPU spent by the task in virtual machine
244 (running a virtual processor).
245 %CPU
247 Total percentage of CPU time used by the task. In an SMP
248 environment, the task's CPU usage will be divided by the
249 total number of CPU's if option -I has been entered on
250 the command line.
251 CPU
253 Processor number to which the task is attached.
254 Command
256 The command name of the task.
257 When reporting global statistics for tasks and all their chil‐
259 dren, the following values may be displayed:
260 UID
262 The real user identification number of the task which is
263 being monitored together with its children.
264 USER
266 The name of the real user owning the task which is being
267 monitored together with its children.
268 PID
270 The identification number of the task which is being mon‐
271 itored together with its children.
272 usr-ms
274 Total number of milliseconds spent by the task and all
275 its children while executing at the user level (applica‐
276 tion), with or without nice priority, and collected dur‐
277 ing the interval of time. Note that this field does NOT
278 include time spent running a virtual processor.
279 system-ms
281 Total number of milliseconds spent by the task and all
282 its children while executing at the system level (ker‐
283 nel), and collected during the interval of time.
284 guest-ms
286 Total number of milliseconds spent by the task and all
287 its children in virtual machine (running a virtual pro‐
288 cessor).
289 Command
291 The command name of the task which is being monitored
292 together with its children.
293 -V Print version number then exit.
295 -w Report task switching activity (kernels 2.6.23 and later only).
297 The following values may be displayed:
298 UID
300 The real user identification number of the task being
301 monitored.
302 USER
304 The name of the real user owning the task being moni‐
305 tored.
306 PID
308 The identification number of the task being monitored.
309 cswch/s
311 Total number of voluntary context switches the task made
312 per second. A voluntary context switch occurs when a
313 task blocks because it requires a resource that is
314 unavailable.
315 nvcswch/s
317 Total number of non voluntary context switches the task
318 made per second. A involuntary context switch takes
319 place when a task executes for the duration of its time
320 slice and then is forced to relinquish the processor.
321 Command
323 The command name of the task.
324
326 The pidstat command takes into account the following environment vari‐
327 able:
328 S_TIME_FORMAT
331 If this variable exists and its value is ISO then the current
332 locale will be ignored when printing the date in the report
333 header. The pidstat command will use the ISO 8601 format (YYYY-
334 MM-DD) instead.
335
338 pidstat 2 5
339 Display five reports of CPU statistics for every active task in
340 the system at two second intervals.
341 pidstat -r -p 1643 2 5
343 Display five reports of page faults and memory statistics for
344 PID 1643 at two second intervals.
345 pidstat -C "fox|bird" -r -p ALL
347 Display global page faults and memory statistics for all the
348 processes whose command name includes the string "fox" or
349 "bird".
350 pidstat -T CHILD -r 2 5
352 Display five reports of page faults statistics at two second
353 intervals for the child processes of all tasks in the system.
354 Only child processes with non-zero statistics values are dis‐
355 played.
356
358 /proc filesystem must be mounted for the pidstat command to work.
359
362 /proc contains various files with system statistics.
363
366 Sebastien Godard (sysstat <at> orange.fr)
367
369 sar(1), top(1), ps(1), mpstat(1), iostat(1), vmstat(8)
370 http://pagesperso-orange.fr/sebastien.godard/
372Linux MARCH 2013 PIDSTAT(1)