1SAR(1) Linux User's Manual SAR(1)
2
3
4
6 sar - Collect, report, or save system activity information.
7
8
10 sar [ -A ] [ -B ] [ -b ] [ -C ] [ -D ] [ -d ] [ -F [ MOUNT ] ] [ -H ] [
11 -h ] [ -p ] [ -r [ ALL ] ] [ -S ] [ -t ] [ -u [ ALL ] ] [ -V ] [ -v ] [
12 -W ] [ -w ] [ -y ] [ -z ] [ --dec={ 0 | 1 | 2 } ] [ --dev=dev_list ] [
13 --fs=fs_list ] [ --help ] [ --human ] [ --iface=iface_list ] [ --pretty
14 ] [ --sadc ] [ -I { int_list | SUM | ALL } ] [ -P { cpu_list | ALL } ]
15 [ -m { keyword[,...] | ALL } ] [ -n { keyword[,...] | ALL } ] [ -q [
16 keyword[,...] | ALL ] ] [ -j { SID | ID | LABEL | PATH | UUID | ... } ]
17 [ -f [ filename ] | -o [ filename ] | -[0-9]+ ] [ -i interval ] [ -s [
18 hh:mm[:ss] ] ] [ -e [ hh:mm[:ss] ] ] [ interval [ count ] ]
19
20
22 The sar command writes to standard output the contents of selected cu‐
23 mulative activity counters in the operating system. The accounting sys‐
24 tem, based on the values in the count and interval parameters, writes
25 information the specified number of times spaced at the specified in‐
26 tervals in seconds. If the interval parameter is set to zero, the sar
27 command displays the average statistics for the time since the system
28 was started. If the interval parameter is specified without the count
29 parameter, then reports are generated continuously. The collected data
30 can also be saved in the file specified by the -o filename flag, in ad‐
31 dition to being displayed onto the screen. If filename is omitted, sar
32 uses the standard system activity daily data file (see below). By de‐
33 fault all the data available from the kernel are saved in the data
34 file.
35
36 The sar command extracts and writes to standard output records previ‐
37 ously saved in a file. This file can be either the one specified by the
38 -f flag or, by default, the standard system activity daily data file.
39 It is also possible to enter -1, -2 etc. as an argument to sar to dis‐
40 play data of that days ago. For example, -1 will point at the standard
41 system activity file of yesterday.
42
43 Standard system activity daily data files are named saDD or saYYYYMMDD,
44 where YYYY stands for the current year, MM for the current month and DD
45 for the current day. They are the default files used by sar only when
46 no filename has been explicitly specified. When used to write data to
47 files (with its option -o), sar will use saYYYYMMDD if option -D has
48 also been specified, else it will use saDD. When used to display the
49 records previously saved in a file, sar will look for the most recent
50 of saDD and saYYYYMMDD, and use it.
51
52 Standard system activity daily data files are located in the
53 /var/log/sa directory by default. Yet it is possible to specify an al‐
54 ternate location for them: If a directory (instead of a plain file) is
55 used with options -f or -o then it will be considered as the directory
56 containing the data files.
57
58 Without the -P flag, the sar command reports system-wide (global among
59 all processors) statistics, which are calculated as averages for values
60 expressed as percentages, and as sums otherwise. If the -P flag is
61 given, the sar command reports activity which relates to the specified
62 processor or processors. If -P ALL is given, the sar command reports
63 statistics for each individual processor and global statistics among
64 all processors. Offline processors are not displayed.
65
66 You can select information about specific system activities using
67 flags. Not specifying any flags selects only CPU activity. Specifying
68 the -A flag selects all possible activities.
69
70 The default version of the sar command (CPU utilization report) might
71 be one of the first facilities the user runs to begin system activity
72 investigation, because it monitors major system resources. If CPU uti‐
73 lization is near 100 percent (user + nice + system), the workload sam‐
74 pled is CPU-bound.
75
76 If multiple samples and multiple reports are desired, it is convenient
77 to specify an output file for the sar command. Run the sar command as a
78 background process. The syntax for this is:
79
80 sar -o datafile interval count >/dev/null 2>&1 &
81
82 All data are captured in binary form and saved to a file (datafile).
83 The data can then be selectively displayed with the sar command using
84 the -f option. Set the interval and count parameters to select count
85 records at interval second intervals. If the count parameter is not
86 set, all the records saved in the file will be selected. Collection of
87 data in this manner is useful to characterize system usage over a pe‐
88 riod of time and determine peak usage hours.
89
90 Note: The sar command only reports on local activities.
91
92
94 -A This is equivalent to specifying -bBdFHSvwWy -I SUM -m ALL -n
95 ALL -q ALL -r ALL -u ALL. This option also implies specifying
96 -I ALL -P ALL unless these options are explicitly set on the
97 command line.
98
99 -B Report paging statistics. The following values are displayed:
100
101 pgpgin/s
102 Total number of kilobytes the system paged in from disk
103 per second.
104
105 pgpgout/s
106 Total number of kilobytes the system paged out to disk
107 per second.
108
109 fault/s
110 Number of page faults (major + minor) made by the system
111 per second. This is not a count of page faults that gen‐
112 erate I/O, because some page faults can be resolved with‐
113 out I/O.
114
115 majflt/s
116 Number of major faults the system has made per second,
117 those which have required loading a memory page from
118 disk.
119
120 pgfree/s
121 Number of pages placed on the free list by the system per
122 second.
123
124 pgscank/s
125 Number of pages scanned by the kswapd daemon per second.
126
127 pgscand/s
128 Number of pages scanned directly per second.
129
130 pgsteal/s
131 Number of pages the system has reclaimed from cache
132 (pagecache and swapcache) per second to satisfy its mem‐
133 ory demands.
134
135 %vmeff Calculated as pgsteal / pgscan, this is a metric of the
136 efficiency of page reclaim. If it is near 100% then al‐
137 most every page coming off the tail of the inactive list
138 is being reaped. If it gets too low (e.g. less than 30%)
139 then the virtual memory is having some difficulty. This
140 field is displayed as zero if no pages have been scanned
141 during the interval of time.
142
143 -b Report I/O and transfer rate statistics. The following values
144 are displayed:
145
146 tps Total number of transfers per second that were issued to
147 physical devices. A transfer is an I/O request to a
148 physical device. Multiple logical requests can be com‐
149 bined into a single I/O request to the device. A trans‐
150 fer is of indeterminate size.
151
152 rtps Total number of read requests per second issued to physi‐
153 cal devices.
154
155 wtps Total number of write requests per second issued to phys‐
156 ical devices.
157
158 dtps Total number of discard requests per second issued to
159 physical devices.
160
161 bread/s
162 Total amount of data read from the devices in blocks per
163 second. Blocks are equivalent to sectors and therefore
164 have a size of 512 bytes.
165
166 bwrtn/s
167 Total amount of data written to devices in blocks per
168 second.
169
170 bdscd/s
171 Total amount of data discarded for devices in blocks per
172 second.
173
174 -C When reading data from a file, tell sar to display comments that
175 have been inserted by sadc.
176
177 -D Use saYYYYMMDD instead of saDD as the standard system activity
178 daily data file name. This option works only when used in con‐
179 junction with option -o to save data to file.
180
181 -d Report activity for each block device. When data are displayed,
182 the device name is displayed as it (should) appear in /dev. sar
183 uses data in /sys to determine the device name based on its ma‐
184 jor and minor numbers. If this name resolution fails, sar will
185 use name mapping controlled by /etc/sysconfig/sysstat.ioconf
186 file. Persistent device names can also be printed if option -j
187 is used (see below). Statistics for all devices are displayed
188 unless a restricted list is specified using option --dev= (see
189 corresponding option entry). Note that disk activity depends on
190 sadc's options -S DISK and -S XDISK to be collected. The follow‐
191 ing values are displayed:
192
193 tps Total number of transfers per second that were issued to
194 physical devices. A transfer is an I/O request to a
195 physical device. Multiple logical requests can be com‐
196 bined into a single I/O request to the device. A trans‐
197 fer is of indeterminate size.
198
199 rkB/s Number of kilobytes read from the device per second.
200
201 wkB/s Number of kilobytes written to the device per second.
202
203 dkB/s Number of kilobytes discarded for the device per second.
204
205 areq-sz
206 The average size (in kilobytes) of the I/O requests that
207 were issued to the device.
208 Note: In previous versions, this field was known as av‐
209 grq-sz and was expressed in sectors.
210
211 aqu-sz The average queue length of the requests that were issued
212 to the device.
213 Note: In previous versions, this field was known as
214 avgqu-sz.
215
216 await The average time (in milliseconds) for I/O requests is‐
217 sued to the device to be served. This includes the time
218 spent by the requests in queue and the time spent servic‐
219 ing them.
220
221 %util Percentage of elapsed time during which I/O requests were
222 issued to the device (bandwidth utilization for the de‐
223 vice). Device saturation occurs when this value is close
224 to 100% for devices serving requests serially. But for
225 devices serving requests in parallel, such as RAID arrays
226 and modern SSDs, this number does not reflect their per‐
227 formance limits.
228
229 --dec={ 0 | 1 | 2 }
230 Specify the number of decimal places to use (0 to 2, default
231 value is 2).
232
233 --dev=dev_list
234 Specify the block devices for which statistics are to be dis‐
235 played by sar. dev_list is a list of comma-separated device
236 names.
237
238 -e [ hh:mm[:ss] ]
239 Set the ending time of the report. The default ending time is
240 18:00:00. Hours must be given in 24-hour format. This option
241 can be used when data are read from or written to a file (op‐
242 tions -f or -o).
243
244 -F [ MOUNT ]
245 Display statistics for currently mounted filesystems. Pseudo-
246 filesystems are ignored. At the end of the report, sar will dis‐
247 play a summary of all those filesystems. Use of the MOUNT param‐
248 eter keyword indicates that mountpoint will be reported instead
249 of filesystem device. Statistics for all filesystems are dis‐
250 played unless a restricted list is specified using option --fs=
251 (see corresponding option entry). Note that filesystems statis‐
252 tics depend on sadc's option -S XDISK to be collected.
253
254 The following values are displayed:
255
256 MBfsfree
257 Total amount of free space in megabytes (including space
258 available only to privileged user).
259
260 MBfsused
261 Total amount of space used in megabytes.
262
263 %fsused
264 Percentage of filesystem space used, as seen by a privi‐
265 leged user.
266
267 %ufsused
268 Percentage of filesystem space used, as seen by an un‐
269 privileged user.
270
271 Ifree Total number of free file nodes in filesystem.
272
273 Iused Total number of file nodes used in filesystem.
274
275 %Iused Percentage of file nodes used in filesystem.
276
277 -f [ filename ]
278 Extract records from filename (created by the -o filename flag).
279 The default value of the filename parameter is the current stan‐
280 dard system activity daily data file. If filename is a directory
281 instead of a plain file then it is considered as the directory
282 where the standard system activity daily data files are located.
283 Option -f is exclusive of option -o.
284
285 --fs=fs_list
286 Specify the filesystems for which statistics are to be displayed
287 by sar. fs_list is a list of comma-separated filesystem names
288 or mountpoints.
289
290 -H Report hugepages utilization statistics. The following values
291 are displayed:
292
293 kbhugfree
294 Amount of hugepages memory in kilobytes that is not yet
295 allocated.
296
297 kbhugused
298 Amount of hugepages memory in kilobytes that has been al‐
299 located.
300
301 %hugused
302 Percentage of total hugepages memory that has been allo‐
303 cated.
304
305 kbhugrsvd
306 Amount of reserved hugepages memory in kilobytes.
307
308 kbhugsurp
309 Amount of surplus hugepages memory in kilobytes.
310
311 -h This option is equivalent to specifying --pretty --human.
312
313 --help Display a short help message then exit.
314
315 --human
316 Print sizes in human readable format (e.g. 1.0k, 1.2M, etc.)
317 The units displayed with this option supersede any other default
318 units (e.g. kilobytes, sectors...) associated with the metrics.
319
320 -I { int_list | SUM | ALL }
321 Report statistics for interrupts. int_list is a list of comma-
322 separated values or range of values (e.g., 0-16,35,400-). The
323 SUM keyword indicates that the total number of interrupts re‐
324 ceived per second is to be displayed. The ALL keyword indicates
325 that statistics from all interrupts, including potential APIC
326 interrupt sources, are to be reported. Note that interrupt sta‐
327 tistics depend on sadc's option -S INT to be collected.
328
329 -i interval
330 Select data records at seconds as close as possible to the num‐
331 ber specified by the interval parameter.
332
333 --iface=iface_list
334 Specify the network interfaces for which statistics are to be
335 displayed by sar. iface_list is a list of comma-separated in‐
336 terface names.
337
338 -j { SID | ID | LABEL | PATH | UUID | ... }
339 Display persistent device names. Use this option in conjunction
340 with option -d. Keywords ID, LABEL, etc. specify the type of the
341 persistent name. These keywords are not limited, only prerequi‐
342 site is that directory with required persistent names is present
343 in /dev/disk. Keyword SID tries to get a stable identifier to
344 use as the device name. A stable identifier won't change across
345 reboots for the same physical device. If it exists, this identi‐
346 fier is normally the WWN (World Wide Name) of the device, as
347 read from the /dev/disk/by-id directory.
348
349 -m { keyword[,...] | ALL }
350 Report power management statistics. Note that these statistics
351 depend on sadc's option -S POWER to be collected.
352
353 Possible keywords are CPU, FAN, FREQ, IN, TEMP and USB.
354
355 With the CPU keyword, statistics about CPU are reported. The
356 following value is displayed:
357
358 MHz Instantaneous CPU clock frequency in MHz.
359
360
361 With the FAN keyword, statistics about fans speed are reported.
362 The following values are displayed:
363
364 rpm Fan speed expressed in revolutions per minute.
365
366 drpm This field is calculated as the difference between cur‐
367 rent fan speed (rpm) and its low limit (fan_min).
368
369 DEVICE Sensor device name.
370
371
372 With the FREQ keyword, statistics about CPU clock frequency are
373 reported. The following value is displayed:
374
375 wghMHz Weighted average CPU clock frequency in MHz. Note that
376 the cpufreq-stats driver must be compiled in the kernel
377 for this option to work.
378
379
380 With the IN keyword, statistics about voltage inputs are re‐
381 ported. The following values are displayed:
382
383 inV Voltage input expressed in Volts.
384
385 %in Relative input value. A value of 100% means that voltage
386 input has reached its high limit (in_max) whereas a value
387 of 0% means that it has reached its low limit (in_min).
388
389 DEVICE Sensor device name.
390
391
392 With the TEMP keyword, statistics about devices temperature are
393 reported. The following values are displayed:
394
395 degC Device temperature expressed in degrees Celsius.
396
397 %temp Relative device temperature. A value of 100% means that
398 temperature has reached its high limit (temp_max).
399
400 DEVICE Sensor device name.
401
402
403 With the USB keyword, the sar command takes a snapshot of all
404 the USB devices currently plugged into the system. At the end of
405 the report, sar will display a summary of all those USB devices.
406 The following values are displayed:
407
408 BUS Root hub number of the USB device.
409
410 idvendor
411 Vendor ID number (assigned by USB organization).
412
413 idprod Product ID number (assigned by Manufacturer).
414
415 maxpower
416 Maximum power consumption of the device (expressed in
417 mA).
418
419 manufact
420 Manufacturer name.
421
422 product
423 Product name.
424
425
426 The ALL keyword is equivalent to specifying all the keywords
427 above and therefore all the power management statistics are re‐
428 ported.
429
430 -n { keyword[,...] | ALL }
431 Report network statistics.
432
433 Possible keywords are DEV, EDEV, FC, ICMP, EICMP, ICMP6, EICMP6,
434 IP, EIP, IP6, EIP6, NFS, NFSD, SOCK, SOCK6, SOFT, TCP, ETCP, UDP
435 and UDP6.
436
437 With the DEV keyword, statistics from the network devices are
438 reported. Statistics for all network interfaces are displayed
439 unless a restricted list is specified using option --iface= (see
440 corresponding option entry). The following values are dis‐
441 played:
442
443 IFACE Name of the network interface for which statistics are
444 reported.
445
446 rxpck/s
447 Total number of packets received per second.
448
449 txpck/s
450 Total number of packets transmitted per second.
451
452 rxkB/s Total number of kilobytes received per second.
453
454 txkB/s Total number of kilobytes transmitted per second.
455
456 rxcmp/s
457 Number of compressed packets received per second (for
458 cslip etc.).
459
460 txcmp/s
461 Number of compressed packets transmitted per second.
462
463 rxmcst/s
464 Number of multicast packets received per second.
465
466 %ifutil
467 Utilization percentage of the network interface. For
468 half-duplex interfaces, utilization is calculated using
469 the sum of rxkB/s and txkB/s as a percentage of the in‐
470 terface speed. For full-duplex, this is the greater of
471 rxkB/S or txkB/s.
472
473
474 With the EDEV keyword, statistics on failures (errors) from the
475 network devices are reported. Statistics for all network inter‐
476 faces are displayed unless a restricted list is specified using
477 option --iface= (see corresponding option entry). The following
478 values are displayed:
479
480 IFACE Name of the network interface for which statistics are
481 reported.
482
483 rxerr/s
484 Total number of bad packets received per second.
485
486 txerr/s
487 Total number of errors that happened per second while
488 transmitting packets.
489
490 coll/s Number of collisions that happened per second while
491 transmitting packets.
492
493 rxdrop/s
494 Number of received packets dropped per second because of
495 a lack of space in linux buffers.
496
497 txdrop/s
498 Number of transmitted packets dropped per second because
499 of a lack of space in linux buffers.
500
501 txcarr/s
502 Number of carrier-errors that happened per second while
503 transmitting packets.
504
505 rxfram/s
506 Number of frame alignment errors that happened per second
507 on received packets.
508
509 rxfifo/s
510 Number of FIFO overrun errors that happened per second on
511 received packets.
512
513 txfifo/s
514 Number of FIFO overrun errors that happened per second on
515 transmitted packets.
516
517
518 With the FC keyword, statistics about fibre channel traffic are
519 reported. Note that fibre channel statistics depend on sadc's
520 option -S DISK to be collected. The following values are dis‐
521 played:
522
523 FCHOST Name of the fibre channel host bus adapter (HBA) inter‐
524 face for which statistics are reported.
525
526 fch_rxf/s
527 The total number of frames received per second.
528
529 fch_txf/s
530 The total number of frames transmitted per second.
531
532 fch_rxw/s
533 The total number of transmission words received per sec‐
534 ond.
535
536 fch_txw/s
537 The total number of transmission words transmitted per
538 second.
539
540
541 With the ICMP keyword, statistics about ICMPv4 network traffic
542 are reported. Note that ICMPv4 statistics depend on sadc's op‐
543 tion -S SNMP to be collected. The following values are dis‐
544 played (formal SNMP names between square brackets):
545
546 imsg/s The total number of ICMP messages which the entity re‐
547 ceived per second [icmpInMsgs]. Note that this counter
548 includes all those counted by ierr/s.
549
550 omsg/s The total number of ICMP messages which this entity at‐
551 tempted to send per second [icmpOutMsgs]. Note that this
552 counter includes all those counted by oerr/s.
553
554 iech/s The number of ICMP Echo (request) messages received per
555 second [icmpInEchos].
556
557 iechr/s
558 The number of ICMP Echo Reply messages received per sec‐
559 ond [icmpInEchoReps].
560
561 oech/s The number of ICMP Echo (request) messages sent per sec‐
562 ond [icmpOutEchos].
563
564 oechr/s
565 The number of ICMP Echo Reply messages sent per second
566 [icmpOutEchoReps].
567
568 itm/s The number of ICMP Timestamp (request) messages received
569 per second [icmpInTimestamps].
570
571 itmr/s The number of ICMP Timestamp Reply messages received per
572 second [icmpInTimestampReps].
573
574 otm/s The number of ICMP Timestamp (request) messages sent per
575 second [icmpOutTimestamps].
576
577 otmr/s The number of ICMP Timestamp Reply messages sent per sec‐
578 ond [icmpOutTimestampReps].
579
580 iadrmk/s
581 The number of ICMP Address Mask Request messages received
582 per second [icmpInAddrMasks].
583
584 iadrmkr/s
585 The number of ICMP Address Mask Reply messages received
586 per second [icmpInAddrMaskReps].
587
588 oadrmk/s
589 The number of ICMP Address Mask Request messages sent per
590 second [icmpOutAddrMasks].
591
592 oadrmkr/s
593 The number of ICMP Address Mask Reply messages sent per
594 second [icmpOutAddrMaskReps].
595
596
597 With the EICMP keyword, statistics about ICMPv4 error messages
598 are reported. Note that ICMPv4 statistics depend on sadc's op‐
599 tion -S SNMP to be collected. The following values are dis‐
600 played (formal SNMP names between square brackets):
601
602 ierr/s The number of ICMP messages per second which the entity
603 received but determined as having ICMP-specific errors
604 (bad ICMP checksums, bad length, etc.) [icmpInErrors].
605
606 oerr/s The number of ICMP messages per second which this entity
607 did not send due to problems discovered within ICMP such
608 as a lack of buffers [icmpOutErrors].
609
610 idstunr/s
611 The number of ICMP Destination Unreachable messages re‐
612 ceived per second [icmpInDestUnreachs].
613
614 odstunr/s
615 The number of ICMP Destination Unreachable messages sent
616 per second [icmpOutDestUnreachs].
617
618 itmex/s
619 The number of ICMP Time Exceeded messages received per
620 second [icmpInTimeExcds].
621
622 otmex/s
623 The number of ICMP Time Exceeded messages sent per second
624 [icmpOutTimeExcds].
625
626 iparmpb/s
627 The number of ICMP Parameter Problem messages received
628 per second [icmpInParmProbs].
629
630 oparmpb/s
631 The number of ICMP Parameter Problem messages sent per
632 second [icmpOutParmProbs].
633
634 isrcq/s
635 The number of ICMP Source Quench messages received per
636 second [icmpInSrcQuenchs].
637
638 osrcq/s
639 The number of ICMP Source Quench messages sent per second
640 [icmpOutSrcQuenchs].
641
642 iredir/s
643 The number of ICMP Redirect messages received per second
644 [icmpInRedirects].
645
646 oredir/s
647 The number of ICMP Redirect messages sent per second
648 [icmpOutRedirects].
649
650
651 With the ICMP6 keyword, statistics about ICMPv6 network traffic
652 are reported. Note that ICMPv6 statistics depend on sadc's op‐
653 tion -S IPV6 to be collected. The following values are dis‐
654 played (formal SNMP names between square brackets):
655
656 imsg6/s
657 The total number of ICMP messages received by the inter‐
658 face per second which includes all those counted by
659 ierr6/s [ipv6IfIcmpInMsgs].
660
661 omsg6/s
662 The total number of ICMP messages which this interface
663 attempted to send per second [ipv6IfIcmpOutMsgs].
664
665 iech6/s
666 The number of ICMP Echo (request) messages received by
667 the interface per second [ipv6IfIcmpInEchos].
668
669 iechr6/s
670 The number of ICMP Echo Reply messages received by the
671 interface per second [ipv6IfIcmpInEchoReplies].
672
673 oechr6/s
674 The number of ICMP Echo Reply messages sent by the inter‐
675 face per second [ipv6IfIcmpOutEchoReplies].
676
677 igmbq6/s
678 The number of ICMPv6 Group Membership Query messages re‐
679 ceived by the interface per second [ipv6IfIcmpInGroupMem‐
680 bQueries].
681
682 igmbr6/s
683 The number of ICMPv6 Group Membership Response messages
684 received by the interface per second [ipv6IfIcmpInGroup‐
685 MembResponses].
686
687 ogmbr6/s
688 The number of ICMPv6 Group Membership Response messages
689 sent per second [ipv6IfIcmpOutGroupMembResponses].
690
691 igmbrd6/s
692 The number of ICMPv6 Group Membership Reduction messages
693 received by the interface per second [ipv6IfIcmpInGroup‐
694 MembReductions].
695
696 ogmbrd6/s
697 The number of ICMPv6 Group Membership Reduction messages
698 sent per second [ipv6IfIcmpOutGroupMembReductions].
699
700 irtsol6/s
701 The number of ICMP Router Solicit messages received by
702 the interface per second [ipv6IfIcmpInRouterSolicits].
703
704 ortsol6/s
705 The number of ICMP Router Solicitation messages sent by
706 the interface per second [ipv6IfIcmpOutRouterSolicits].
707
708 irtad6/s
709 The number of ICMP Router Advertisement messages received
710 by the interface per second [ipv6IfIcmpInRouterAdvertise‐
711 ments].
712
713 inbsol6/s
714 The number of ICMP Neighbor Solicit messages received by
715 the interface per second [ipv6IfIcmpInNeighborSolicits].
716
717 onbsol6/s
718 The number of ICMP Neighbor Solicitation messages sent by
719 the interface per second [ipv6IfIcmpOutNeighborSolicits].
720
721 inbad6/s
722 The number of ICMP Neighbor Advertisement messages re‐
723 ceived by the interface per second [ipv6IfIcmpInNeighbo‐
724 rAdvertisements].
725
726 onbad6/s
727 The number of ICMP Neighbor Advertisement messages sent
728 by the interface per second [ipv6IfIcmpOutNeighborAdver‐
729 tisements].
730
731
732 With the EICMP6 keyword, statistics about ICMPv6 error messages
733 are reported. Note that ICMPv6 statistics depend on sadc's op‐
734 tion -S IPV6 to be collected. The following values are dis‐
735 played (formal SNMP names between square brackets):
736
737 ierr6/s
738 The number of ICMP messages per second which the inter‐
739 face received but determined as having ICMP-specific er‐
740 rors (bad ICMP checksums, bad length, etc.)
741 [ipv6IfIcmpInErrors]
742
743 idtunr6/s
744 The number of ICMP Destination Unreachable messages re‐
745 ceived by the interface per second [ipv6IfIcmpInDestUn‐
746 reachs].
747
748 odtunr6/s
749 The number of ICMP Destination Unreachable messages sent
750 by the interface per second [ipv6IfIcmpOutDestUnreachs].
751
752 itmex6/s
753 The number of ICMP Time Exceeded messages received by the
754 interface per second [ipv6IfIcmpInTimeExcds].
755
756 otmex6/s
757 The number of ICMP Time Exceeded messages sent by the in‐
758 terface per second [ipv6IfIcmpOutTimeExcds].
759
760 iprmpb6/s
761 The number of ICMP Parameter Problem messages received by
762 the interface per second [ipv6IfIcmpInParmProblems].
763
764 oprmpb6/s
765 The number of ICMP Parameter Problem messages sent by the
766 interface per second [ipv6IfIcmpOutParmProblems].
767
768 iredir6/s
769 The number of Redirect messages received by the interface
770 per second [ipv6IfIcmpInRedirects].
771
772 oredir6/s
773 The number of Redirect messages sent by the interface by
774 second [ipv6IfIcmpOutRedirects].
775
776 ipck2b6/s
777 The number of ICMP Packet Too Big messages received by
778 the interface per second [ipv6IfIcmpInPktTooBigs].
779
780 opck2b6/s
781 The number of ICMP Packet Too Big messages sent by the
782 interface per second [ipv6IfIcmpOutPktTooBigs].
783
784
785 With the IP keyword, statistics about IPv4 network traffic are
786 reported. Note that IPv4 statistics depend on sadc's option -S
787 SNMP to be collected. The following values are displayed (for‐
788 mal SNMP names between square brackets):
789
790 irec/s The total number of input datagrams received from inter‐
791 faces per second, including those received in error [ip‐
792 InReceives].
793
794 fwddgm/s
795 The number of input datagrams per second, for which this
796 entity was not their final IP destination, as a result of
797 which an attempt was made to find a route to forward them
798 to that final destination [ipForwDatagrams].
799
800 idel/s The total number of input datagrams successfully deliv‐
801 ered per second to IP user-protocols (including ICMP)
802 [ipInDelivers].
803
804 orq/s The total number of IP datagrams which local IP user-pro‐
805 tocols (including ICMP) supplied per second to IP in re‐
806 quests for transmission [ipOutRequests]. Note that this
807 counter does not include any datagrams counted in fwd‐
808 dgm/s.
809
810 asmrq/s
811 The number of IP fragments received per second which
812 needed to be reassembled at this entity [ipReasmReqds].
813
814 asmok/s
815 The number of IP datagrams successfully re-assembled per
816 second [ipReasmOKs].
817
818 fragok/s
819 The number of IP datagrams that have been successfully
820 fragmented at this entity per second [ipFragOKs].
821
822 fragcrt/s
823 The number of IP datagram fragments that have been gener‐
824 ated per second as a result of fragmentation at this en‐
825 tity [ipFragCreates].
826
827
828 With the EIP keyword, statistics about IPv4 network errors are
829 reported. Note that IPv4 statistics depend on sadc's option -S
830 SNMP to be collected. The following values are displayed (for‐
831 mal SNMP names between square brackets):
832
833 ihdrerr/s
834 The number of input datagrams discarded per second due to
835 errors in their IP headers, including bad checksums, ver‐
836 sion number mismatch, other format errors, time-to-live
837 exceeded, errors discovered in processing their IP op‐
838 tions, etc. [ipInHdrErrors]
839
840 iadrerr/s
841 The number of input datagrams discarded per second be‐
842 cause the IP address in their IP header's destination
843 field was not a valid address to be received at this en‐
844 tity. This count includes invalid addresses (e.g.,
845 0.0.0.0) and addresses of unsupported Classes (e.g.,
846 Class E). For entities which are not IP routers and
847 therefore do not forward datagrams, this counter includes
848 datagrams discarded because the destination address was
849 not a local address [ipInAddrErrors].
850
851 iukwnpr/s
852 The number of locally-addressed datagrams received suc‐
853 cessfully but discarded per second because of an unknown
854 or unsupported protocol [ipInUnknownProtos].
855
856 idisc/s
857 The number of input IP datagrams per second for which no
858 problems were encountered to prevent their continued pro‐
859 cessing, but which were discarded (e.g., for lack of buf‐
860 fer space) [ipInDiscards]. Note that this counter does
861 not include any datagrams discarded while awaiting re-as‐
862 sembly.
863
864 odisc/s
865 The number of output IP datagrams per second for which no
866 problem was encountered to prevent their transmission to
867 their destination, but which were discarded (e.g., for
868 lack of buffer space) [ipOutDiscards]. Note that this
869 counter would include datagrams counted in fwddgm/s if
870 any such packets met this (discretionary) discard crite‐
871 rion.
872
873 onort/s
874 The number of IP datagrams discarded per second because
875 no route could be found to transmit them to their desti‐
876 nation [ipOutNoRoutes]. Note that this counter includes
877 any packets counted in fwddgm/s which meet this 'no-
878 route' criterion. Note that this includes any datagrams
879 which a host cannot route because all of its default
880 routers are down.
881
882 asmf/s The number of failures detected per second by the IP re-
883 assembly algorithm (for whatever reason: timed out, er‐
884 rors, etc) [ipReasmFails]. Note that this is not neces‐
885 sarily a count of discarded IP fragments since some algo‐
886 rithms can lose track of the number of fragments by com‐
887 bining them as they are received.
888
889 fragf/s
890 The number of IP datagrams that have been discarded per
891 second because they needed to be fragmented at this en‐
892 tity but could not be, e.g., because their Don't Fragment
893 flag was set [ipFragFails].
894
895
896 With the IP6 keyword, statistics about IPv6 network traffic are
897 reported. Note that IPv6 statistics depend on sadc's option -S
898 IPV6 to be collected. The following values are displayed (for‐
899 mal SNMP names between square brackets):
900
901 irec6/s
902 The total number of input datagrams received from inter‐
903 faces per second, including those received in error
904 [ipv6IfStatsInReceives].
905
906 fwddgm6/s
907 The number of output datagrams per second which this en‐
908 tity received and forwarded to their final destinations
909 [ipv6IfStatsOutForwDatagrams].
910
911 idel6/s
912 The total number of datagrams successfully delivered per
913 second to IPv6 user-protocols (including ICMP) [ipv6IfS‐
914 tatsInDelivers].
915
916 orq6/s The total number of IPv6 datagrams which local IPv6 user-
917 protocols (including ICMP) supplied per second to IPv6 in
918 requests for transmission [ipv6IfStatsOutRequests]. Note
919 that this counter does not include any datagrams counted
920 in fwddgm6/s.
921
922 asmrq6/s
923 The number of IPv6 fragments received per second which
924 needed to be reassembled at this interface [ipv6IfStat‐
925 sReasmReqds].
926
927 asmok6/s
928 The number of IPv6 datagrams successfully reassembled per
929 second [ipv6IfStatsReasmOKs].
930
931 imcpck6/s
932 The number of multicast packets received per second by
933 the interface [ipv6IfStatsInMcastPkts].
934
935 omcpck6/s
936 The number of multicast packets transmitted per second by
937 the interface [ipv6IfStatsOutMcastPkts].
938
939 fragok6/s
940 The number of IPv6 datagrams that have been successfully
941 fragmented at this output interface per second [ipv6IfS‐
942 tatsOutFragOKs].
943
944 fragcr6/s
945 The number of output datagram fragments that have been
946 generated per second as a result of fragmentation at this
947 output interface [ipv6IfStatsOutFragCreates].
948
949
950 With the EIP6 keyword, statistics about IPv6 network errors are
951 reported. Note that IPv6 statistics depend on sadc's option -S
952 IPV6 to be collected. The following values are displayed (for‐
953 mal SNMP names between square brackets):
954
955 ihdrer6/s
956 The number of input datagrams discarded per second due to
957 errors in their IPv6 headers, including version number
958 mismatch, other format errors, hop count exceeded, errors
959 discovered in processing their IPv6 options, etc.
960 [ipv6IfStatsInHdrErrors]
961
962 iadrer6/s
963 The number of input datagrams discarded per second be‐
964 cause the IPv6 address in their IPv6 header's destination
965 field was not a valid address to be received at this en‐
966 tity. This count includes invalid addresses (e.g., ::0)
967 and unsupported addresses (e.g., addresses with unallo‐
968 cated prefixes). For entities which are not IPv6 routers
969 and therefore do not forward datagrams, this counter in‐
970 cludes datagrams discarded because the destination ad‐
971 dress was not a local address [ipv6IfStatsInAddrErrors].
972
973 iukwnp6/s
974 The number of locally-addressed datagrams received suc‐
975 cessfully but discarded per second because of an unknown
976 or unsupported protocol [ipv6IfStatsInUnknownProtos].
977
978 i2big6/s
979 The number of input datagrams that could not be forwarded
980 per second because their size exceeded the link MTU of
981 outgoing interface [ipv6IfStatsInTooBigErrors].
982
983 idisc6/s
984 The number of input IPv6 datagrams per second for which
985 no problems were encountered to prevent their continued
986 processing, but which were discarded (e.g., for lack of
987 buffer space) [ipv6IfStatsInDiscards]. Note that this
988 counter does not include any datagrams discarded while
989 awaiting re-assembly.
990
991 odisc6/s
992 The number of output IPv6 datagrams per second for which
993 no problem was encountered to prevent their transmission
994 to their destination, but which were discarded (e.g., for
995 lack of buffer space) [ipv6IfStatsOutDiscards]. Note that
996 this counter would include datagrams counted in fwddgm6/s
997 if any such packets met this (discretionary) discard cri‐
998 terion.
999
1000 inort6/s
1001 The number of input datagrams discarded per second be‐
1002 cause no route could be found to transmit them to their
1003 destination [ipv6IfStatsInNoRoutes].
1004
1005 onort6/s
1006 The number of locally generated IP datagrams discarded
1007 per second because no route could be found to transmit
1008 them to their destination [unknown formal SNMP name].
1009
1010 asmf6/s
1011 The number of failures detected per second by the IPv6
1012 re-assembly algorithm (for whatever reason: timed out,
1013 errors, etc.) [ipv6IfStatsReasmFails]. Note that this is
1014 not necessarily a count of discarded IPv6 fragments since
1015 some algorithms can lose track of the number of fragments
1016 by combining them as they are received.
1017
1018 fragf6/s
1019 The number of IPv6 datagrams that have been discarded per
1020 second because they needed to be fragmented at this out‐
1021 put interface but could not be [ipv6IfStatsOutFragFails].
1022
1023 itrpck6/s
1024 The number of input datagrams discarded per second be‐
1025 cause datagram frame didn't carry enough data [ipv6IfS‐
1026 tatsInTruncatedPkts].
1027
1028
1029 With the NFS keyword, statistics about NFS client activity are
1030 reported. The following values are displayed:
1031
1032 call/s Number of RPC requests made per second.
1033
1034 retrans/s
1035 Number of RPC requests per second, those which needed to
1036 be retransmitted (for example because of a server time‐
1037 out).
1038
1039 read/s Number of 'read' RPC calls made per second.
1040
1041 write/s
1042 Number of 'write' RPC calls made per second.
1043
1044 access/s
1045 Number of 'access' RPC calls made per second.
1046
1047 getatt/s
1048 Number of 'getattr' RPC calls made per second.
1049
1050
1051 With the NFSD keyword, statistics about NFS server activity are
1052 reported. The following values are displayed:
1053
1054 scall/s
1055 Number of RPC requests received per second.
1056
1057 badcall/s
1058 Number of bad RPC requests received per second, those
1059 whose processing generated an error.
1060
1061 packet/s
1062 Number of network packets received per second.
1063
1064 udp/s Number of UDP packets received per second.
1065
1066 tcp/s Number of TCP packets received per second.
1067
1068 hit/s Number of reply cache hits per second.
1069
1070 miss/s Number of reply cache misses per second.
1071
1072 sread/s
1073 Number of 'read' RPC calls received per second.
1074
1075 swrite/s
1076 Number of 'write' RPC calls received per second.
1077
1078 saccess/s
1079 Number of 'access' RPC calls received per second.
1080
1081 sgetatt/s
1082 Number of 'getattr' RPC calls received per second.
1083
1084
1085 With the SOCK keyword, statistics on sockets in use are reported
1086 (IPv4). The following values are displayed:
1087
1088 totsck Total number of sockets used by the system.
1089
1090 tcpsck Number of TCP sockets currently in use.
1091
1092 udpsck Number of UDP sockets currently in use.
1093
1094 rawsck Number of RAW sockets currently in use.
1095
1096 ip-frag
1097 Number of IP fragments currently in queue.
1098
1099 tcp-tw Number of TCP sockets in TIME_WAIT state.
1100
1101
1102 With the SOCK6 keyword, statistics on sockets in use are re‐
1103 ported (IPv6). Note that IPv6 statistics depend on sadc's op‐
1104 tion -S IPV6 to be collected. The following values are dis‐
1105 played:
1106
1107 tcp6sck
1108 Number of TCPv6 sockets currently in use.
1109
1110 udp6sck
1111 Number of UDPv6 sockets currently in use.
1112
1113 raw6sck
1114 Number of RAWv6 sockets currently in use.
1115
1116 ip6-frag
1117 Number of IPv6 fragments currently in use.
1118
1119
1120 With the SOFT keyword, statistics about software-based network
1121 processing are reported. The following values are displayed:
1122
1123 total/s
1124 The total number of network frames processed per second.
1125
1126 dropd/s
1127 The total number of network frames dropped per second be‐
1128 cause there was no room on the processing queue.
1129
1130 squeezd/s
1131 The number of times the softirq handler function termi‐
1132 nated per second because its budget was consumed or the
1133 time limit was reached, but more work could have been
1134 done.
1135
1136 rx_rps/s
1137 The number of times the CPU has been woken up per second
1138 to process packets via an inter-processor interrupt.
1139
1140 flw_lim/s
1141 The number of times the flow limit has been reached per
1142 second. Flow limiting is an optional RPS feature that
1143 can be used to limit the number of packets queued to the
1144 backlog for each flow to a certain amount. This can help
1145 ensure that smaller flows are processed even though much
1146 larger flows are pushing packets in.
1147
1148
1149 With the TCP keyword, statistics about TCPv4 network traffic are
1150 reported. Note that TCPv4 statistics depend on sadc's option -S
1151 SNMP to be collected. The following values are displayed (for‐
1152 mal SNMP names between square brackets):
1153
1154 active/s
1155 The number of times TCP connections have made a direct
1156 transition to the SYN-SENT state from the CLOSED state
1157 per second [tcpActiveOpens].
1158
1159 passive/s
1160 The number of times TCP connections have made a direct
1161 transition to the SYN-RCVD state from the LISTEN state
1162 per second [tcpPassiveOpens].
1163
1164 iseg/s The total number of segments received per second, includ‐
1165 ing those received in error [tcpInSegs]. This count in‐
1166 cludes segments received on currently established connec‐
1167 tions.
1168
1169 oseg/s The total number of segments sent per second, including
1170 those on current connections but excluding those contain‐
1171 ing only retransmitted octets [tcpOutSegs].
1172
1173
1174 With the ETCP keyword, statistics about TCPv4 network errors are
1175 reported. Note that TCPv4 statistics depend on sadc's option -S
1176 SNMP to be collected. The following values are displayed (for‐
1177 mal SNMP names between square brackets):
1178
1179 atmptf/s
1180 The number of times per second TCP connections have made
1181 a direct transition to the CLOSED state from either the
1182 SYN-SENT state or the SYN-RCVD state, plus the number of
1183 times per second TCP connections have made a direct tran‐
1184 sition to the LISTEN state from the SYN-RCVD state [tc‐
1185 pAttemptFails].
1186
1187 estres/s
1188 The number of times per second TCP connections have made
1189 a direct transition to the CLOSED state from either the
1190 ESTABLISHED state or the CLOSE-WAIT state [tcpEstabRe‐
1191 sets].
1192
1193 retrans/s
1194 The total number of segments retransmitted per second -
1195 that is, the number of TCP segments transmitted contain‐
1196 ing one or more previously transmitted octets [tcpRe‐
1197 transSegs].
1198
1199 isegerr/s
1200 The total number of segments received in error (e.g., bad
1201 TCP checksums) per second [tcpInErrs].
1202
1203 orsts/s
1204 The number of TCP segments sent per second containing the
1205 RST flag [tcpOutRsts].
1206
1207
1208 With the UDP keyword, statistics about UDPv4 network traffic are
1209 reported. Note that UDPv4 statistics depend on sadc's option -S
1210 SNMP to be collected. The following values are displayed (for‐
1211 mal SNMP names between square brackets):
1212
1213 idgm/s The total number of UDP datagrams delivered per second to
1214 UDP users [udpInDatagrams].
1215
1216 odgm/s The total number of UDP datagrams sent per second from
1217 this entity [udpOutDatagrams].
1218
1219 noport/s
1220 The total number of received UDP datagrams per second for
1221 which there was no application at the destination port
1222 [udpNoPorts].
1223
1224 idgmerr/s
1225 The number of received UDP datagrams per second that
1226 could not be delivered for reasons other than the lack of
1227 an application at the destination port [udpInErrors].
1228
1229
1230 With the UDP6 keyword, statistics about UDPv6 network traffic
1231 are reported. Note that UDPv6 statistics depend on sadc's op‐
1232 tion -S IPV6 to be collected. The following values are dis‐
1233 played (formal SNMP names between square brackets):
1234
1235 idgm6/s
1236 The total number of UDP datagrams delivered per second to
1237 UDP users [udpInDatagrams].
1238
1239 odgm6/s
1240 The total number of UDP datagrams sent per second from
1241 this entity [udpOutDatagrams].
1242
1243 noport6/s
1244 The total number of received UDP datagrams per second for
1245 which there was no application at the destination port
1246 [udpNoPorts].
1247
1248 idgmer6/s
1249 The number of received UDP datagrams per second that
1250 could not be delivered for reasons other than the lack of
1251 an application at the destination port [udpInErrors].
1252
1253
1254 The ALL keyword is equivalent to specifying all the keywords
1255 above and therefore all the network activities are reported.
1256
1257 -o [ filename ]
1258 Save the readings in the file in binary form. Each reading is in
1259 a separate record. The default value of the filename parameter
1260 is the current standard system activity daily data file. If
1261 filename is a directory instead of a plain file then it is con‐
1262 sidered as the directory where the standard system activity
1263 daily data files are located. Option -o is exclusive of option
1264 -f. All the data available from the kernel are saved in the
1265 file (in fact, sar calls its data collector sadc with the option
1266 -S ALL. See sadc(8) manual page).
1267
1268 -P { cpu_list | ALL }
1269 Report per-processor statistics for the specified processor or
1270 processors. cpu_list is a list of comma-separated values or
1271 range of values (e.g., 0,2,4-7,12-). Note that processor 0 is
1272 the first processor, and processor all is the global average
1273 among all processors. Specifying the ALL keyword reports sta‐
1274 tistics for each individual processor, and globally for all pro‐
1275 cessors. Offline processors are not displayed.
1276
1277 -p, --pretty
1278 Make reports easier to read by a human. This option may be es‐
1279 pecially useful when displaying e.g., network interfaces or
1280 block devices statistics.
1281
1282 -q [ keyword[,...] | ALL ]
1283 Report system load and pressure-stall statistics.
1284
1285 Possible keywords are CPU, IO, LOAD, MEM and PSI".
1286
1287 With the CPU keyword, CPU pressure statistics are reported. The
1288 following values are displayed:
1289
1290 %scpu-10
1291 Percentage of the time that at least some runnable tasks
1292 were delayed because the CPU was unavailable to them,
1293 over the last 10 second window.
1294
1295 %scpu-60
1296 Percentage of the time that at least some runnable tasks
1297 were delayed because the CPU was unavailable to them,
1298 over the last 60 second window.
1299
1300 %scpu-300
1301 Percentage of the time that at least some runnable tasks
1302 were delayed because the CPU was unavailable to them,
1303 over the last 300 second window.
1304
1305 %scpu Percentage of the time that at least some runnable tasks
1306 were delayed because the CPU was unavailable to them,
1307 over the last time interval.
1308
1309
1310 With the IO keyword, I/O pressure statistics are reported. The
1311 following values are displayed:
1312
1313 %sio-10
1314 Percentage of the time that at least some tasks lost
1315 waiting for I/O, over the last 10 second window.
1316
1317 %sio-60
1318 Percentage of the time that at least some tasks lost
1319 waiting for I/O, over the last 60 second window.
1320
1321 %sio-300
1322 Percentage of the time that at least some tasks lost
1323 waiting for I/O, over the last 300 second window.
1324
1325 %sio Percentage of the time that at least some tasks lost
1326 waiting for I/O, over the last time interval.
1327
1328 %fio-10
1329 Percentage of the time during which all non-idle tasks
1330 were stalled waiting for I/O, over the last 10 second
1331 window.
1332
1333 %fio-60
1334 Percentage of the time during which all non-idle tasks
1335 were stalled waiting for I/O, over the last 60 second
1336 window.
1337
1338 %fio-300
1339 Percentage of the time during which all non-idle tasks
1340 were stalled waiting for I/O, over the last 300 second
1341 window.
1342
1343 %fio Percentage of the time during which all non-idle tasks
1344 were stalled waiting for I/O, over the last time inter‐
1345 val.
1346
1347
1348 With the LOAD keyword, queue length and load averages statistics
1349 are reported. The following values are displayed:
1350
1351 runq-sz
1352 Run queue length (number of tasks running or waiting for
1353 run time).
1354
1355 plist-sz
1356 Number of tasks in the task list.
1357
1358 ldavg-1
1359 System load average for the last minute. The load aver‐
1360 age is calculated as the average number of runnable or
1361 running tasks (R state), and the number of tasks in unin‐
1362 terruptible sleep (D state) over the specified interval.
1363
1364 ldavg-5
1365 System load average for the past 5 minutes.
1366
1367 ldavg-15
1368 System load average for the past 15 minutes.
1369
1370 blocked
1371 Number of tasks currently blocked, waiting for I/O to
1372 complete.
1373
1374
1375 With the MEM keyword, memory pressure statistics are reported.
1376 The following values are displayed:
1377
1378 %smem-10
1379 Percentage of the time during which at least some tasks
1380 were waiting for memory resources, over the last 10 sec‐
1381 ond window.
1382
1383 %smem-60
1384 Percentage of the time during which at least some tasks
1385 were waiting for memory resources, over the last 60 sec‐
1386 ond window.
1387
1388 %smem-300
1389 Percentage of the time during which at least some tasks
1390 were waiting for memory resources, over the last 300 sec‐
1391 ond window.
1392
1393 %smem Percentage of the time during which at least some tasks
1394 were waiting for memory resources, over the last time in‐
1395 terval.
1396
1397 %fmem-10
1398 Percentage of the time during which all non-idle tasks
1399 were stalled waiting for memory resources, over the last
1400 10 second window.
1401
1402 %fmem-60
1403 Percentage of the time during which all non-idle tasks
1404 were stalled waiting for memory resources, over the last
1405 60 second window.
1406
1407 %fmem-300
1408 Percentage of the time during which all non-idle tasks
1409 were stalled waiting for memory resources, over the last
1410 300 second window.
1411
1412 %fmem Percentage of the time during which all non-idle tasks
1413 were stalled waiting for memory resources, over the last
1414 time interval.
1415
1416
1417 The PSI keyword is equivalent to specifying CPU, IO and MEM key‐
1418 words together and therefore all the pressure-stall statistics
1419 are reported.
1420
1421 The ALL keyword is equivalent to specifying all the keywords
1422 above and therefore all the statistics are reported.
1423
1424 -r [ ALL ]
1425 Report memory utilization statistics. The ALL keyword indicates
1426 that all the memory fields should be displayed. The following
1427 values may be displayed:
1428
1429 kbmemfree
1430 Amount of free memory available in kilobytes.
1431
1432 kbavail
1433 Estimate of how much memory in kilobytes is available for
1434 starting new applications, without swapping. The esti‐
1435 mate takes into account that the system needs some page
1436 cache to function well, and that not all reclaimable slab
1437 will be reclaimable, due to items being in use. The im‐
1438 pact of those factors will vary from system to system.
1439
1440 kbmemused
1441 Amount of used memory in kilobytes (calculated as total
1442 installed memory - kbmemfree - kbbuffers - kbcached - kb‐
1443 slab).
1444
1445 %memused
1446 Percentage of used memory.
1447
1448 kbbuffers
1449 Amount of memory used as buffers by the kernel in kilo‐
1450 bytes.
1451
1452 kbcached
1453 Amount of memory used to cache data by the kernel in
1454 kilobytes.
1455
1456 kbcommit
1457 Amount of memory in kilobytes needed for current work‐
1458 load. This is an estimate of how much RAM/swap is needed
1459 to guarantee that there never is out of memory.
1460
1461 %commit
1462 Percentage of memory needed for current workload in rela‐
1463 tion to the total amount of memory (RAM+swap). This num‐
1464 ber may be greater than 100% because the kernel usually
1465 overcommits memory.
1466
1467 kbactive
1468 Amount of active memory in kilobytes (memory that has
1469 been used more recently and usually not reclaimed unless
1470 absolutely necessary).
1471
1472 kbinact
1473 Amount of inactive memory in kilobytes (memory which has
1474 been less recently used. It is more eligible to be re‐
1475 claimed for other purposes).
1476
1477 kbdirty
1478 Amount of memory in kilobytes waiting to get written back
1479 to the disk.
1480
1481 kbanonpg
1482 Amount of non-file backed pages in kilobytes mapped into
1483 userspace page tables.
1484
1485 kbslab Amount of memory in kilobytes used by the kernel to cache
1486 data structures for its own use.
1487
1488 kbkstack
1489 Amount of memory in kilobytes used for kernel stack
1490 space.
1491
1492 kbpgtbl
1493 Amount of memory in kilobytes dedicated to the lowest
1494 level of page tables.
1495
1496 kbvmused
1497 Amount of memory in kilobytes of used virtual address
1498 space.
1499
1500 -S Report swap space utilization statistics. The following values
1501 are displayed:
1502
1503 kbswpfree
1504 Amount of free swap space in kilobytes.
1505
1506 kbswpused
1507 Amount of used swap space in kilobytes.
1508
1509 %swpused
1510 Percentage of used swap space.
1511
1512 kbswpcad
1513 Amount of cached swap memory in kilobytes. This is mem‐
1514 ory that once was swapped out, is swapped back in but
1515 still also is in the swap area (if memory is needed it
1516 doesn't need to be swapped out again because it is al‐
1517 ready in the swap area. This saves I/O).
1518
1519 %swpcad
1520 Percentage of cached swap memory in relation to the
1521 amount of used swap space.
1522
1523 -s [ hh:mm[:ss] ]
1524 Set the starting time of the data, causing the sar command to
1525 extract records time-tagged at, or following, the time speci‐
1526 fied. The default starting time is 08:00:00. Hours must be
1527 given in 24-hour format. This option can be used only when data
1528 are read from a file (option -f).
1529
1530 --sadc Indicate which data collector is called by sar. If the data
1531 collector is sought in PATH then enter "which sadc" to know
1532 where it is located.
1533
1534 -t When reading data from a daily data file, indicate that sar
1535 should display the timestamps in the original local time of the
1536 data file creator. Without this option, the sar command displays
1537 the timestamps in the user's locale time.
1538
1539 -u [ ALL ]
1540 Report CPU utilization. The ALL keyword indicates that all the
1541 CPU fields should be displayed. The report may show the follow‐
1542 ing fields:
1543
1544 %user Percentage of CPU utilization that occurred while execut‐
1545 ing at the user level (application). Note that this field
1546 includes time spent running virtual processors.
1547
1548 %usr Percentage of CPU utilization that occurred while execut‐
1549 ing at the user level (application). Note that this field
1550 does NOT include time spent running virtual processors.
1551
1552 %nice Percentage of CPU utilization that occurred while execut‐
1553 ing at the user level with nice priority.
1554
1555 %system
1556 Percentage of CPU utilization that occurred while execut‐
1557 ing at the system level (kernel). Note that this field
1558 includes time spent servicing hardware and software in‐
1559 terrupts.
1560
1561 %sys Percentage of CPU utilization that occurred while execut‐
1562 ing at the system level (kernel). Note that this field
1563 does NOT include time spent servicing hardware or soft‐
1564 ware interrupts.
1565
1566 %iowait
1567 Percentage of time that the CPU or CPUs were idle during
1568 which the system had an outstanding disk I/O request.
1569
1570 %steal Percentage of time spent in involuntary wait by the vir‐
1571 tual CPU or CPUs while the hypervisor was servicing an‐
1572 other virtual processor.
1573
1574 %irq Percentage of time spent by the CPU or CPUs to service
1575 hardware interrupts.
1576
1577 %soft Percentage of time spent by the CPU or CPUs to service
1578 software interrupts.
1579
1580 %guest Percentage of time spent by the CPU or CPUs to run a vir‐
1581 tual processor.
1582
1583 %gnice Percentage of time spent by the CPU or CPUs to run a
1584 niced guest.
1585
1586 %idle Percentage of time that the CPU or CPUs were idle and the
1587 system did not have an outstanding disk I/O request.
1588
1589 -V Print version number then exit.
1590
1591 -v Report status of inode, file and other kernel tables. The fol‐
1592 lowing values are displayed:
1593
1594 dentunusd
1595 Number of unused cache entries in the directory cache.
1596
1597 file-nr
1598 Number of file handles used by the system.
1599
1600 inode-nr
1601 Number of inode handlers used by the system.
1602
1603 pty-nr Number of pseudo-terminals used by the system.
1604
1605 -W Report swapping statistics. The following values are displayed:
1606
1607 pswpin/s
1608 Total number of swap pages the system brought in per sec‐
1609 ond.
1610
1611 pswpout/s
1612 Total number of swap pages the system brought out per
1613 second.
1614
1615 -w Report task creation and system switching activity. The follow‐
1616 ing values are displayed:
1617
1618 proc/s Total number of tasks created per second.
1619
1620 cswch/s
1621 Total number of context switches per second.
1622
1623 -y Report TTY devices activity. The following values are displayed:
1624
1625 rcvin/s
1626 Number of receive interrupts per second for current se‐
1627 rial line. Serial line number is given in the TTY col‐
1628 umn.
1629
1630 xmtin/s
1631 Number of transmit interrupts per second for current se‐
1632 rial line.
1633
1634 framerr/s
1635 Number of frame errors per second for current serial
1636 line.
1637
1638 prtyerr/s
1639 Number of parity errors per second for current serial
1640 line.
1641
1642 brk/s Number of breaks per second for current serial line.
1643
1644 ovrun/s
1645 Number of overrun errors per second for current serial
1646 line.
1647
1648 -z Tell sar to omit output for any devices for which there was no
1649 activity during the sample period.
1650
1651
1653 The sar command takes into account the following environment variables:
1654
1655 S_COLORS
1656 By default statistics are displayed in color when the output is
1657 connected to a terminal. Use this variable to change the set‐
1658 tings. Possible values for this variable are never, always or
1659 auto (the latter is equivalent to the default settings).
1660 Please note that the color (being red, yellow, or some other
1661 color) used to display a value is not indicative of any kind of
1662 issue simply because of the color. It only indicates different
1663 ranges of values.
1664
1665 S_COLORS_SGR
1666 Specify the colors and other attributes used to display statis‐
1667 tics on the terminal. Its value is a colon-separated list of
1668 capabilities that defaults to
1669 C=33;22:H=31;1:I=32;22:M=35;1:N=34;1:R=31;22:Z=34;22. Supported
1670 capabilities are:
1671
1672 C= SGR (Select Graphic Rendition) substring for comments in‐
1673 serted in the binary daily data files.
1674
1675 H= SGR substring for percentage values greater than or equal
1676 to 75%.
1677
1678 I= SGR substring for item names or values (eg. network in‐
1679 terfaces, CPU number...)
1680
1681 M= SGR substring for percentage values in the range from 50%
1682 to 75%.
1683
1684 N= SGR substring for non-zero statistics values.
1685
1686 R= SGR substring for restart messages.
1687
1688 Z= SGR substring for zero values.
1689
1690 S_TIME_DEF_TIME
1691 If this variable exists and its value is UTC then sar will save
1692 its data in UTC time (data will still be displayed in local
1693 time). sar will also use UTC time instead of local time to de‐
1694 termine the current daily data file located in the /var/log/sa
1695 directory. This variable may be useful for servers with users
1696 located across several timezones.
1697
1698 S_TIME_FORMAT
1699 If this variable exists and its value is ISO then the current
1700 locale will be ignored when printing the date in the report
1701 header. The sar command will use the ISO 8601 format (YYYY-MM-
1702 DD) instead. The timestamp will also be compliant with ISO 8601
1703 format.
1704
1705
1707 sar -u 2 5
1708 Report CPU utilization for each 2 seconds. 5 lines are dis‐
1709 played.
1710
1711 sar -I 14 -o int14.file 2 10
1712 Report statistics on IRQ 14 for each 2 seconds. 10 lines are
1713 displayed. Data are stored in a file called int14.file.
1714
1715 sar -r -n DEV -f /var/log/sa/sa16
1716 Display memory and network statistics saved in daily data file
1717 sa16.
1718
1719 sar -A Display all the statistics saved in current daily data file.
1720
1721
1723 /proc filesystem must be mounted for the sar command to work.
1724
1725 All the statistics are not necessarily available, depending on the ker‐
1726 nel version used. sar assumes that you are using at least a 2.6 ker‐
1727 nel.
1728
1729 Although sar speaks of kilobytes (kB), megabytes (MB)..., it actually
1730 uses kibibytes (kiB), mebibytes (MiB)... A kibibyte is equal to 1024
1731 bytes, and a mebibyte is equal to 1024 kibibytes.
1732
1733
1735 /var/log/sa/saDD
1736 /var/log/sa/saYYYYMMDD
1737 The standard system activity daily data files and their default
1738 location. YYYY stands for the current year, MM for the current
1739 month and DD for the current day.
1740
1741 /proc and /sys contain various files with system statistics.
1742
1743
1745 Sebastien Godard (sysstat <at> orange.fr)
1746
1747
1749 sadc(8), sa1(8), sa2(8), sadf(1), sysstat(5), pidstat(1), mpstat(1),
1750 iostat(1), vmstat(8)
1751
1752 https://github.com/sysstat/sysstat
1753 http://pagesperso-orange.fr/sebastien.godard/
1754
1755
1756
1757Linux AUGUST 2020 SAR(1)