Print this page
NEX-3729 KRRP changes mess up iostat(1M)
Reviewed by: Alek Pinchuk <alek.pinchuk@nexenta.com>
Reviewed by: Roman Strashkin <roman.strashkin@nexenta.com>
| Split |
Close |
| Expand all |
| Collapse all |
--- old/usr/src/man/man1m/iostat.1m.man.txt
+++ new/usr/src/man/man1m/iostat.1m.man.txt
1 1 IOSTAT(1M) Maintenance Commands IOSTAT(1M)
2 2
3 3
4 4
5 5 NAME
6 6 iostat - report I/O statistics
7 7
8 8 SYNOPSIS
9 - /usr/bin/iostat [-cCdDeEiImMnpPrstxXYz] [-l n] [-T u | d]
9 + /usr/bin/iostat [-cCdDeEiImMnpPfFrstxXYz] [-l n] [-T u | d]
10 10 [disk]... [interval [count]]
11 11
12 12
13 13 DESCRIPTION
14 14 The iostat utility iteratively reports terminal, disk, and tape I/O
15 15 activity, as well as CPU utilization. The first line of output is for
16 16 all time since boot; each subsequent line is for the prior interval
17 17 only.
18 18
19 19
20 20 To compute this information, the kernel maintains a number of counters.
21 21 For each disk, the kernel counts reads, writes, bytes read, and bytes
22 22 written. The kernel also takes hi-res time stamps at queue entry and
23 23 exit points, which allows it to keep track of the residence time and
24 24 cumulative residence-length product for each queue. Using these values,
25 25 iostat produces highly accurate measures of throughput, utilization,
26 26 queue lengths, transaction rates and service time. For terminals
27 27 collectively, the kernel simply counts the number of input and output
28 28 characters.
29 29
30 30
31 31 During execution of the kernel status command, the state of the system
32 32 can change. If relevant, a state change message is included in the
33 33 iostat output, in one of the following forms:
34 34
35 35 <<device added: sd0>>
36 36 <<device removed: sd0>>
37 37 <<partition added: sd0,a>>
38 38 <<partition removed: sd0,a>>
39 39 <<NFS mounted: nfs1>>
40 40 <<NFS unmounted: nfs1>>
41 41 <<multi-path added: ssd4>>
42 42 <<multi-path removed: ssd4>>
43 43 <<controller added: c1>>
44 44 <<controller removed: c1>>
45 45 <<processors added: 1, 3>>
46 46 <<processors removed: 1, 3>>
47 47
48 48
49 49
50 50
51 51 Note that the names printed in these state change messages are affected
52 52 by the -n and -m options as appropriate.
53 53
54 54
55 55 For more general system statistics, use sar(1), sar(1M), or vmstat(1M).
56 56
57 57 Output
58 58 The output of the iostat utility includes the following information.
59 59
60 60 device
61 61 name of the disk
62 62
63 63
64 64 r/s
65 65 reads per second
66 66
67 67
68 68 w/s
69 69 writes per second
70 70
71 71
72 72 kr/s
73 73 kilobytes read per second
74 74
75 75 The average I/O size during the interval can be computed from
76 76 kr/s divided by r/s.
77 77
78 78
79 79 kw/s
80 80 kilobytes written per second
81 81
82 82 The average I/O size during the interval can be computed from
83 83 kw/s divided by w/s.
84 84
85 85
86 86 wait
87 87 average number of transactions waiting for service (queue
88 88 length)
89 89
90 90 This is the number of I/O operations held in the device
91 91 driver queue waiting for acceptance by the device.
92 92
93 93
94 94 actv
95 95 average number of transactions actively being serviced
96 96 (removed from the queue but not yet completed)
97 97
98 98 This is the number of I/O operations accepted, but not yet
99 99 serviced, by the device.
100 100
101 101
102 102 svc_t
103 103 average response time of transactions, in milliseconds
104 104
105 105 The svc_t output reports the overall response time, rather
106 106 than the service time, of a device. The overall time includes
107 107 the time that transactions are in queue and the time that
108 108 transactions are being serviced. The time spent in queue is
109 109 shown with the -x option in the wsvc_t output column. The
110 110 time spent servicing transactions is the true service time.
111 111 Service time is also shown with the -x option and appears in
112 112 the asvc_t output column of the same report.
113 113
114 114
115 115 %w
116 116 percent of time there are transactions waiting for service
117 117 (queue non-empty)
118 118
119 119
120 120 %b
121 121 percent of time the disk is busy (transactions in progress)
122 122
123 123
124 124 wsvc_t
125 125 average service time in wait queue, in milliseconds
126 126
127 127
128 128 asvc_t
129 129 average service time of active transactions, in milliseconds
130 130
131 131
132 132 OPTIONS
133 133 The following options are supported:
134 134
135 135 -c
136 136 Report the percentage of time the system has spent in user
137 137 mode, in system mode, in dtrace probes, and idling. See the
138 138 NOTES section and mpstat(1m) for more information.
139 139
140 140
141 141 -C
142 142 When the -x option is also selected, report extended disk
143 143 statistics aggregated by controller id.
144 144
145 145
146 146 -d
147 147 For each disk, report the number of kilobytes transferred
148 148 per second, the number of transfers per second, and the
149 149 average service time in milliseconds.
150 150
151 151
152 152 -D
153 153 For each disk, report the reads per second, writes per
154 154 second, and percentage disk utilization.
155 155
156 156
157 157 -e
158 158 Display device error summary statistics. The total errors,
159 159 hard errors, soft errors, and transport errors are
160 160 displayed.
161 161
162 162
163 163 -E
164 164 Display all device error statistics.
165 165
166 166
167 167 -i
168 168 In -E output, display the Device ID instead of the Serial
169 169 No. The Device Id is a unique identifier registered by a
170 170 driver through ddi_devid_register(9F).
171 171
172 172
173 173 -I
174 174 Report the counts in each interval, rather than rates
175 175 (where applicable).
176 176
177 177
178 178 -l n
179 179 Limit the number of disks included in the report to n; the
180 180 disk limit defaults to 4 for -d and -D, and unlimited for
181 181 -x. Note: disks explicitly requested (see disk below) are
182 182 not subject to this disk limit.
183 183
184 184
185 185 -m
186 186 Report file system mount points. This option is most useful
187 187 if the -P or -p option is also specified or used in
188 188 conjunction with -Xn or -en. The -m option is useful only
189 189 if the mount point is actually listed in the output. This
190 190 option can only be used in conjunction with the -n option.
191 191
192 192
193 193 -M
194 194 Display data throughput in MB/sec instead of KB/sec.
195 195
196 196
197 197 -n
198 198 Display names in descriptive format. For example, cXtYdZ,
199 199 rmt/N, server:/export/path.
200 200
201 201 By default, disks are identified by instance names such as
202 202 ssd23 or md301. Combining the -n option with the -x option
203 203 causes disk names to display in the cXtYdZsN format which
204 204 is more easily associated with physical hardware
205 205 characteristics. The cXtYdZsN format is particularly useful
206 206 in FibreChannel (FC) environments where the FC World Wide
207 207 Name appears in the t field.
208 208
209 209
|
↓ open down ↓ |
190 lines elided |
↑ open up ↑ |
210 210 -p
211 211 For each disk, report per-partition statistics in addition
212 212 to per-device statistics.
213 213
214 214
215 215 -P
216 216 For each disk, report per-partition statistics only, no
217 217 per-device statistics.
218 218
219 219
220 + -f
221 + Report ZFS-level statistics for ZFS pool and individual
222 + vdevs in addition to per-device statistics.
223 +
224 +
225 + -F
226 + Report ZFS pool and individual physical vdevs statistics
227 + only, no per-device statistics.
228 +
229 +
220 230 -r
221 231 Display data in a comma-separated format.
222 232
223 233
224 234 -s
225 235 Suppress messages related to state changes.
226 236
227 237
228 238 -t
229 239 Report the number of characters read and written to
230 240 terminals per second.
231 241
232 242
233 243 -T u | d
234 244 Display a time stamp.
235 245
236 246 Specify u for a printed representation of the internal
237 247 representation of time. See time(2). Specify d for standard
238 248 date format. See date(1).
239 249
240 250
241 251 -X
242 252 For disks under scsi_vhci(7D) control, in addition to disk
243 253 lun statistics, also report statistics for lun.controller.
244 254
245 255
246 256 -x
247 257 Report extended disk statistics. By default, disks are
248 258 identified by instance names such as ssd23 or md301.
249 259 Combining the x option with the -n option causes disk names
250 260 to display in the cXtYdZsN format, more easily associated
251 261 with physical hardware characteristics. Using the cXtYdZsN
252 262 format is particularly helpful in the FibreChannel
253 263 environments where the FC World Wide Name appears in the t
254 264 field.
255 265
256 266 If no output display is requested (no -x, -e, -E), -x is
257 267 implied.
258 268
259 269
260 270 -Y
261 271 For disks under scsi_vhci(7D) control, in addition to disk
262 272 lun statistics, also report statistics for lun.targetport
263 273 and lun.targetport.controller.
264 274
265 275 In -n (descriptive) mode the targetport is shown in using
266 276 the target-port property of the path. Without -n the
267 277 targetport is shown using the shorter port-id. All target
268 278 ports with the same target-port property value share the
269 279 same port-id. The target-port-to-port-id association does
270 280 not persist across reboot.
271 281
272 282 If no output display is requested (no -x, -e, -E), -x is
273 283 implied.
274 284
275 285
276 286 -z
277 287 Do not print lines whose underlying data values are all
278 288 zeros.
279 289
280 290
281 291
282 292 The option set -xcnCXTdz interval is particularly useful for
283 293 determining whether disk I/O problems exist and for identifying
284 294 problems.
285 295
286 296 OPERANDS
287 297 The following operands are supported:
288 298
289 299 count
290 300 Display only count reports.
291 301
292 302
293 303 disk
294 304 Explicitly specify the disks to be reported; in addition to
295 305 any explicit disks, any active disks up to the disk limit
296 306 (see -l above) will also be reported.
297 307
298 308
299 309 interval
300 310 Report once each interval seconds.
301 311
302 312
303 313 EXAMPLES
304 314 Example 1 Using iostat to Generate User and System Operation Statistics
305 315
306 316
307 317 The following command displays two reports of extended device
308 318 statistics, aggregated by controller id, for user (us) and system (sy)
309 319 operations. Because the -n option is used with the -x option, devices
310 320 are identified by controller names.
311 321
312 322
313 323 example% iostat -xcnCXTdz 5
314 324
315 325
316 326 Mon Nov 24 14:58:36 2003
317 327 cpu
318 328 us sy dt id
319 329 14 31 0 20
320 330 extended device statistics
321 331 r/s w/s kr/s kw wait actv wsvc_t asvc_t %w %b device
322 332 3.8 29.9 145.8 44.0 0.0 0.2 0.1 6.4 0 5 c0
323 333 666.3 814.8 12577.6 17591.1 91.3 82.3 61.6 55.6 0 2 c12
324 334 180.0 234.6 4401.1 5712.6 0.0 147.7 0.0 356.3 0 98 d10
325 335
326 336 Mon Nov 24 14:58:41 2003
327 337 cpu
328 338 us sy dt id
329 339 11 31 1 22
330 340 extended device statistics
331 341 r/s w/s kr/s kw wait actv wsvc_t asvc_t %w %b device
332 342 0.8 41.0 5.2 20.5 0.0 0.2 0.2 4.4 0 6 c0
333 343 565.3 581.7 8573.2 10458.9 0.0 26.6 0.0 23.2 0 3 c12
334 344 106.5 81.3 3393.2 1948.6 0.0 5.7 0.0 30.1 0 99 d10
335 345
336 346
337 347
338 348 Example 2 Using iostat to Generate TTY Statistics
339 349
340 350
341 351 The following command displays two reports on the activity of five
342 352 disks in different modes of operation. Because the -x option is used,
343 353 disks are identified by instance names.
344 354
345 355
346 356 example% iostat -x tc 5 2
347 357
348 358
349 359 extended device statistics tty cpu
350 360 device r/s w/s kr/s kw/s wait actv svc_t %w %b tin tout us sy dt id
351 361 sd0 0.4 0.3 10.4 8.0 0.0 0.0 36.9 0 1 0 10 0 0 0 99
352 362 sd1 0.0 0.0 0.3 0.4 0.0 0.0 35.0 0 0
353 363 sd6 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
354 364 nfs1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
355 365 nfs2 0.0 0.0 0.0 0.1 0.0 0.0 35.6 0 0
356 366 extended device statistics tty cpu
357 367 device r/s w/s kr/s kw/s wait actv svc_t %w %b tin tout us sy dt id
358 368 sd0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 155 0 0 0 100
359 369 sd1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
360 370 sd6 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
361 371 nfs1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
362 372 nfs2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
363 373
364 374
365 375
366 376
367 377 Example 3 Using iostat to Generate Partition and Device Statistics
368 378
369 379
370 380 The following command generates partition and device statistics for
371 381 each disk. Because the -n option is used with the -x option, disks are
372 382 identified by controller names.
373 383
374 384
375 385 example% iostat -xnp
376 386
377 387 extended device statistics
378 388 r/s w/s kr/s kw/s wait actv wsvc_t asvc_t %w %b device
379 389 0.4 0.3 10.4 7.9 0.0 0.0 0.0 36.9 0 1 c0t0d0
380 390 0.3 0.3 9.0 7.3 0.0 0.0 0.0 37.2 0 1 c0t0d0s0
381 391 0.0 0.0 0.1 0.5 0.0 0.0 0.0 34.0 0 0 c0t0d0s1
382 392 0.0 0.0 0.0 0.1 0.0 0.0 0.6 35.0 0 0 fuji:/export/home/user3
383 393
384 394
385 395
386 396 Example 4 Show Translation from Instance Name to Descriptive Name
387 397
388 398
389 399 The following example illustrates the use of iostat to translate a
390 400 specific instance name to a descriptive name.
391 401
392 402
393 403 example% iostat -xn sd1
394 404 extended device statistics
395 405 r/s w/s kr/s kw/s wait actv wsvc_t asvc_t %w %b device
396 406 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 c8t1d0
397 407
398 408
399 409
400 410 Example 5 Show Target Port and Controller Activity for a Specific Disk
401 411
402 412
403 413 In the following example, there are four controllers, all connected to
404 414 the same target port.
405 415
406 416
407 417 # iostat -Y ssd22
408 418 extended device statistics
409 419 device r/s w/s kr/s kw/s wait actv svc_t %w %b
410 420 ssd22 0.2 0.0 1.5 0.0 0.0 0.0 0.7 0 0
411 421 ssd22.t2 0.2 0.0 1.5 0.0 0.0 0.0 0.0 0 0
412 422 ssd22.t2.fp0 0.0 0.0 0.4 0.0 0.0 0.0 0.0 0 0
413 423 ssd22.t2.fp1 0.0 0.0 0.4 0.0 0.0 0.0 0.0 0 0
414 424 ssd22.t2.fp2 0.0 0.0 0.4 0.0 0.0 0.0 0.0 0 0
415 425 ssd22.t2.fp3 0.0 0.0 0.4 0.0 0.0 0.0 0.0 0 0
416 426
417 427
418 428
419 429 ATTRIBUTES
420 430 See attributes(5) for descriptions of the following attributes:
421 431
422 432
423 433
424 434
425 435 +--------------------+-----------------+
426 436 | ATTRIBUTE TYPE | ATTRIBUTE VALUE |
427 437 +--------------------+-----------------+
428 438 |Interface Stability | See below. |
429 439 +--------------------+-----------------+
430 440
431 441
432 442 Invocation is evolving. Human readable output is unstable.
433 443
434 444 SEE ALSO
435 445 date(1), sar(1), sar(1M), mpstat(1M), vmstat(1M), time(2),
436 446 attributes(5), scsi_vhci(7D)
437 447
438 448 NOTES
439 449 The sum of CPU utilization might vary slightly from 100 because of
440 450 rounding errors in the production of a percentage figure.
441 451
442 452
443 453 The svc_t response time is not particularly significant when the I/0
444 454 (r/s+w/s) rates are under 0.5 per second. Harmless spikes are fairly
445 455 normal in such cases.
446 456
447 457
448 458 The mpstat utility reports the same dt, usr, and sys statistics. See
449 459 mpstat(1M) for more information.
450 460
451 461
452 462 When executed in a zone and if the pools facility is active, iostat(1M)
453 463 will only provide information for those processors in the processor set
454 464 of the pool to which the zone is bound.
455 465
456 466
457 467
458 468 March 23, 2009 IOSTAT(1M)
|
↓ open down ↓ |
229 lines elided |
↑ open up ↑ |
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX