1PIDSTAT(1) Linux User's Manual PIDSTAT(1)
2
6 pidstat - Report statistics for Linux tasks.
7
9 pidstat [ -d ] [ -H ] [ -h ] [ -I ] [ -l ] [ -R ] [ -r ] [ -s ] [ -t ]
10 [ -U [ username ] ] [ -u ] [ -V ] [ -v ] [ -w ] [ -C comm ] [ -G
11 process_name ] [ --dec={ 0 | 1 | 2 } ] [ --human ] [ -p { pid [,...] |
12 SELF | ALL } ] [ -T { TASK | CHILD | ALL } ] [ interval [ count ] ] [
13 -e program args ]
14
16 The pidstat command is used for monitoring individual tasks currently
17 being managed by the Linux kernel. It writes to standard output activ‐
18 ities for every task selected with option -p or for every task managed
19 by the Linux kernel if option -p ALL has been used. Not selecting any
20 tasks is equivalent to specifying -p ALL but only active tasks (tasks
21 with non-zero statistics values) will appear in the report.
22 The pidstat command can also be used for monitoring the child processes
24 of selected tasks. Read about option -T below.
25 The interval parameter specifies the amount of time in seconds between
27 each report. A value of 0 (or no parameters at all) indicates that
28 tasks statistics are to be reported for the time since system startup
29 (boot). The count parameter can be specified in conjunction with the
30 interval parameter if this one is not set to zero. The value of count
31 determines the number of reports generated at interval seconds apart.
32 If the interval parameter is specified without the count parameter, the
33 pidstat command generates reports continuously.
34 You can select information about specific task activities using flags.
36 Not specifying any flags selects only CPU activity.
37
40 -C comm
41 Display only tasks whose command name includes the string comm.
42 This string can be a regular expression.
43 -d Report I/O statistics (kernels 2.6.20 and later only). The fol‐
45 lowing values may be displayed:
46 UID
48 The real user identification number of the task being
49 monitored.
50 USER
52 The name of the real user owning the task being moni‐
53 tored.
54 PID
56 The identification number of the task being monitored.
57 kB_rd/s
59 Number of kilobytes the task has caused to be read from
60 disk per second.
61 kB_wr/s
63 Number of kilobytes the task has caused, or shall cause
64 to be written to disk per second.
65 kB_ccwr/s
67 Number of kilobytes whose writing to disk has been can‐
68 celled by the task. This may occur when the task trun‐
69 cates some dirty pagecache. In this case, some IO which
70 another task has been accounted for will not be happen‐
71 ing.
72 iodelay
74 Block I/O delay of the task being monitored, measured in
75 clock ticks. This metric includes the delays spent wait‐
76 ing for sync block I/O completion and for swapin block
77 I/O completion.
78 Command
80 The command name of the task.
81 --dec={ 0 | 1 | 2 }
83 Specify the number of decimal places to use (0 to 2, default
84 value is 2).
85 -e program args
87 Execute program with given arguments args and monitor it with
88 pidstat. pidstat stops when program terminates.
89 -G process_name
91 Display only processes whose command name includes the string
92 process_name. This string can be a regular expression. If
93 option -t is used together with option -G then the threads
94 belonging to that process are also displayed (even if their com‐
95 mand name doesn't include the string process_name).
96 -H Display timestamp in seconds since the epoch.
98 -h Display all activities horizontally on a single line, with no
100 average statistics at the end of the report. This is intended to
101 make it easier to be parsed by other programs.
102 --human
104 Print sizes in human readable format (e.g. 1.0k, 1.2M, etc.)
105 The units displayed with this option supersede any other default
106 units (e.g. kilobytes, sectors...) associated with the metrics.
107 -I In an SMP environment, indicate that tasks CPU usage (as dis‐
109 played by option -u ) should be divided by the total number of
110 processors.
111 -l Display the process command name and all its arguments.
113 -p { pid [,...] | SELF | ALL }
115 Select tasks (processes) for which statistics are to be
116 reported. pid is the process identification number. The SELF
117 keyword indicates that statistics are to be reported for the
118 pidstat process itself, whereas the ALL keyword indicates that
119 statistics are to be reported for all the tasks managed by the
120 system.
121 -R Report realtime priority and scheduling policy information. The
123 following values may be displayed:
124 UID
126 The real user identification number of the task being
127 monitored.
128 USER
130 The name of the real user owning the task being moni‐
131 tored.
132 PID
134 The identification number of the task being monitored.
135 prio
137 The realtime priority of the task being monitored.
138 policy
140 The scheduling policy of the task being monitored.
141 Command
143 The command name of the task.
144 -r Report page faults and memory utilization.
146 When reporting statistics for individual tasks, the following
148 values may be displayed:
149 UID
151 The real user identification number of the task being
152 monitored.
153 USER
155 The name of the real user owning the task being moni‐
156 tored.
157 PID
159 The identification number of the task being monitored.
160 minflt/s
162 Total number of minor faults the task has made per sec‐
163 ond, those which have not required loading a memory page
164 from disk.
165 majflt/s
167 Total number of major faults the task has made per sec‐
168 ond, those which have required loading a memory page from
169 disk.
170 VSZ
172 Virtual Size: The virtual memory usage of entire task in
173 kilobytes.
174 RSS
176 Resident Set Size: The non-swapped physical memory used
177 by the task in kilobytes.
178 %MEM
180 The tasks's currently used share of available physical
181 memory.
182 Command
184 The command name of the task.
185 When reporting global statistics for tasks and all their chil‐
187 dren, the following values may be displayed:
188 UID
190 The real user identification number of the task which is
191 being monitored together with its children.
192 USER
194 The name of the real user owning the task which is being
195 monitored together with its children.
196 PID
198 The identification number of the task which is being mon‐
199 itored together with its children.
200 minflt-nr
202 Total number of minor faults made by the task and all its
203 children, and collected during the interval of time.
204 majflt-nr
206 Total number of major faults made by the task and all its
207 children, and collected during the interval of time.
208 Command
210 The command name of the task which is being monitored
211 together with its children.
212 -s Report stack utilization. The following values may be dis‐
214 played:
215 UID
217 The real user identification number of the task being
218 monitored.
219 USER
221 The name of the real user owning the task being moni‐
222 tored.
223 PID
225 The identification number of the task being monitored.
226 StkSize
228 The amount of memory in kilobytes reserved for the task
229 as stack, but not necessarily used.
230 StkRef
232 The amount of memory in kilobytes used as stack, refer‐
233 enced by the task.
234 Command
236 The command name of the task.
237 -T { TASK | CHILD | ALL }
239 This option specifies what has to be monitored by the pidstat
240 command. The TASK keyword indicates that statistics are to be
241 reported for individual tasks (this is the default option)
242 whereas the CHILD keyword indicates that statistics are to be
243 globally reported for the selected tasks and all their children.
244 The ALL keyword indicates that statistics are to be reported for
245 individual tasks and globally for the selected tasks and their
246 children.
247 Note: Global statistics for tasks and all their children are not
249 available for all options of pidstat. Also these statistics are
250 not necessarily relevant to current time interval: The statis‐
251 tics of a child process are collected only when it finishes or
252 it is killed.
253 -t Also display statistics for threads associated with selected
255 tasks.
256 This option adds the following values to the reports:
258 TGID
260 The identification number of the thread group leader.
261 TID
263 The identification number of the thread being monitored.
264 -U [ username ]
266 Display the real user name of the tasks being monitored instead
267 of the UID. If username is specified, then only tasks belonging
268 to the specified user are displayed.
269 -u Report CPU utilization.
271 When reporting statistics for individual tasks, the following
273 values may be displayed:
274 UID
276 The real user identification number of the task being
277 monitored.
278 USER
280 The name of the real user owning the task being moni‐
281 tored.
282 PID
284 The identification number of the task being monitored.
285 %usr
287 Percentage of CPU used by the task while executing at the
288 user level (application), with or without nice priority.
289 Note that this field does NOT include time spent running
290 a virtual processor.
291 %system
293 Percentage of CPU used by the task while executing at the
294 system level (kernel).
295 %guest
297 Percentage of CPU spent by the task in virtual machine
298 (running a virtual processor).
299 %wait
301 Percentage of CPU spent by the task while waiting to run.
302 %CPU
304 Total percentage of CPU time used by the task. In an SMP
305 environment, the task's CPU usage will be divided by the
306 total number of CPU's if option -I has been entered on
307 the command line.
308 CPU
310 Processor number to which the task is attached.
311 Command
313 The command name of the task.
314 When reporting global statistics for tasks and all their chil‐
316 dren, the following values may be displayed:
317 UID
319 The real user identification number of the task which is
320 being monitored together with its children.
321 USER
323 The name of the real user owning the task which is being
324 monitored together with its children.
325 PID
327 The identification number of the task which is being mon‐
328 itored together with its children.
329 usr-ms
331 Total number of milliseconds spent by the task and all
332 its children while executing at the user level (applica‐
333 tion), with or without nice priority, and collected dur‐
334 ing the interval of time. Note that this field does NOT
335 include time spent running a virtual processor.
336 system-ms
338 Total number of milliseconds spent by the task and all
339 its children while executing at the system level (ker‐
340 nel), and collected during the interval of time.
341 guest-ms
343 Total number of milliseconds spent by the task and all
344 its children in virtual machine (running a virtual pro‐
345 cessor).
346 Command
348 The command name of the task which is being monitored
349 together with its children.
350 -V Print version number then exit.
352 -v Report values of some kernel tables. The following values may be
354 displayed:
355 UID
357 The real user identification number of the task being
358 monitored.
359 USER
361 The name of the real user owning the task being moni‐
362 tored.
363 PID
365 The identification number of the task being monitored.
366 threads
368 Number of threads associated with current task.
369 fd-nr
371 Number of file descriptors associated with current task.
372 Command
374 The command name of the task.
375 -w Report task switching activity (kernels 2.6.23 and later only).
377 The following values may be displayed:
378 UID
380 The real user identification number of the task being
381 monitored.
382 USER
384 The name of the real user owning the task being moni‐
385 tored.
386 PID
388 The identification number of the task being monitored.
389 cswch/s
391 Total number of voluntary context switches the task made
392 per second. A voluntary context switch occurs when a
393 task blocks because it requires a resource that is
394 unavailable.
395 nvcswch/s
397 Total number of non voluntary context switches the task
398 made per second. A involuntary context switch takes
399 place when a task executes for the duration of its time
400 slice and then is forced to relinquish the processor.
401 Command
403 The command name of the task.
404
406 The pidstat command takes into account the following environment vari‐
407 ables:
408 S_COLORS
411 When this variable is set, display statistics in color on the
412 terminal. Possible values for this variable are never, always
413 or auto (the latter is the default).
414 Please note that the color (being red, yellow, or some other
416 color) used to display a value is not indicative of any kind of
417 issue simply because of the color. It only indicates different
418 ranges of values.
419 S_COLORS_SGR
422 Specify the colors and other attributes used to display statis‐
423 tics on the terminal. Its value is a colon-separated list of
424 capabilities that defaults to
425 H=31;1:I=32;22:M=35;1:N=34;1:Z=34;22. Supported capabilities
426 are:
427 H= SGR (Select Graphic Rendition) substring for percentage
430 values greater than or equal to 75%.
431 I= SGR substring for item values like PID, UID or CPU num‐
434 ber.
435 M= SGR substring for percentage values in the range from 50%
438 to 75%.
439 N= SGR substring for non-zero statistics values and for
442 tasks names.
443 Z= SGR substring for zero values and for threads names.
446 S_TIME_FORMAT
449 If this variable exists and its value is ISO then the current
450 locale will be ignored when printing the date in the report
451 header. The pidstat command will use the ISO 8601 format (YYYY-
452 MM-DD) instead. The timestamp will also be compliant with ISO
453 8601 format.
454
457 pidstat 2 5
458 Display five reports of CPU statistics for every active task in
459 the system at two second intervals.
460 pidstat -r -p 1643 2 5
462 Display five reports of page faults and memory statistics for
463 PID 1643 at two second intervals.
464 pidstat -C "fox|bird" -r -p ALL
466 Display global page faults and memory statistics for all the
467 processes whose command name includes the string "fox" or
468 "bird".
469 pidstat -T CHILD -r 2 5
471 Display five reports of page faults statistics at two second
472 intervals for the child processes of all tasks in the system.
473 Only child processes with non-zero statistics values are dis‐
474 played.
475
477 /proc filesystem must be mounted for the pidstat command to work.
478
481 /proc contains various files with system statistics.
482
485 Sebastien Godard (sysstat <at> orange.fr)
486
488 sar(1), top(1), ps(1), mpstat(1), iostat(1), vmstat(8)
489 https://github.com/sysstat/sysstat
491 http://pagesperso-orange.fr/sebastien.godard/
493Linux JULY 2018 PIDSTAT(1)