1 .TH BLKPARSE 1 "March 6, 2007" "blktrace git\-20070306202522" ""
5 blkparse \- produce formatted output of event streams of block devices
9 .B blkparse [ \fIoptions\fR ]
14 The \fIblkparse\fR utility will attempt to combine streams of events for
15 various devices on various CPUs, and produce a formatted output of the event
16 information. Specifically, it will take the (machine-readable) output of the
17 \fIblktrace\fR utility and convert it to a nicely formatted and human-readable
20 As with \fIblktrace\fR, some details concerning \fIblkparse\fR
21 will help in understanding the command line options presented below.
26 By default, \fIblkparse\fR expects to run in a post-processing mode; one where
27 the trace events have been saved by a previous run of blktrace, and blkparse
28 is combining event streams and dumping formatted data.
30 blkparse may be run in a live manner concurrently with blktrace by specifying
31 \fB\-i \-\fR to blkparse, and combining it with the live option for blktrace.
34 % blktrace \-d /dev/sda \-o \- | blkparse \-i \-
38 You can set how many blkparse batches event reads via the \fB\-b\fR option, the
39 default is to handle events in batches of 512.
43 If you have saved event traces in blktrace with different output names (via
44 the \fB\-o\fR option to blktrace), you must specify the same input name via the
49 The format of the output data can be controlled via the \fB\-f\fR or \fB\-F\fR
50 options \-\- see OUTPUT DESCRIPTION AND FORMATTING for details.
53 By default, blkparse sends formatted data to standard output. This may
54 be changed via the \fB\-o\fR option, or text output can be disabled via the
55 \fB\-O\fR option. A merged binary stream can be produced using the \fB\-d\fR
63 \-\-set-mask=\fIhex-mask\fR
65 Set filter mask to \fIhex-mask\fR, see blktrace (8) for masks
70 \-\-act-mask=\fImask\fR
72 Add \fImask\fR to current filter, see blktrace (8) for masks
77 \-\-input-directory=\fIdir\fR
79 Prepend \fIdir\fR to input file names
86 Standard input read batching
93 Specifies base name for input files \-\- default is \fIdevice\fR.blktrace.\fIcpu\fR.
95 As noted above, specifying \fB\-i \-\fR runs in live mode with blktrace
96 (reading data from standard in).
101 \-\-format=\fItyp,fmt\fR
105 \-\-format\-spec=\fIfmt\fR
108 (See OUTPUT DESCRIPTION AND FORMATTING for details.)
110 The \-f form specifies a format for all events
112 The \-F form allows one to specify a format for a specific
113 event type. The single\-character \fItyp\fR field is one of the
114 action specifiers described in ACTION IDENTIFIERS.
121 When \-d is specified, this will stop messages from being output to the
122 file. (Can seriously reduce the size of the resultant file when using
123 the CFQ I/O scheduler.)
130 Hash processes by name, not by PID
135 \-\-output=\fIfile\fR
144 Do \fInot\fR produce text output, used for binary (\fB\-d\fR) only
149 \-\-dump\-binary=\fIfile\fR
163 \-\-per\-program\-stats
165 Displays data sorted by program
172 Display time deltas per IO
177 \-\-stopwatch=\fIspan\fR
179 Display traces for the \fIspan\fR specified \-\- where span can be:
181 \fIend\-time\fR \-\- Display traces from time 0 through \fIend\-time\fR (in ns)
185 \fIstart:end\-time\fR \-\- Display traces from time \fIstart\fR
186 through end\-time (in ns).
193 More verbose marginal on marginal errors
205 The following trace actions are recognised:
209 A previously issued request has been completed. The output will detail the
210 sector and size of that request, as well as the success or failure of it.
214 A request that previously resided on the block layer queue or in the i/o
215 scheduler has been sent to the driver.
219 A request is being sent to the i/o scheduler for addition to the internal queue
220 and later service by the driver. The request is fully formed at this time.
224 This notes intent to queue i/o at the given location. No real requests exists
229 The data pages attached to this \fIbio\fR are not reachable by the hardware
230 and must be bounced to a lower memory location. This causes a big slowdown in
231 i/o performance, since the data must be copied to/from kernel buffers. Usually
232 this can be fixed with using better hardware -- either a better i/o controller,
233 or a platform with an IOMMU.
236 \fBM -- back merge\fR
237 A previously inserted request exists that ends on the boundary of where this i/o
238 begins, so the i/o scheduler can merge them together.
241 \fBF -- front merge\fR
242 Same as the back merge, except this i/o ends where a previously inserted
246 \fBM --front or back merge\fR
250 \fBM -- front or back merge\fR
254 \fBG -- get request\fR
255 To send any type of request to a block device, a \fIstruct request\fR
256 container must be allocated first.
260 No available request structures were available, so the issuer has to wait for
265 When i/o is queued to a previously empty block device queue, Linux will plug the
266 queue in anticipation of future ios being added before this data is needed.
270 Some request data already queued in the device, start sending requests to the
271 driver. This may happen automatically if a timeout period has passed (see next
272 entry) or if a number of requests have been added to the queue.
275 \fBT -- unplug due to timer\fR
276 If nobody requests the i/o that was queued after plugging the queue, Linux will
277 automatically unplug it after a defined period has passed.
281 On raid or device mapper setups, an incoming i/o may straddle a device or
282 internal zone and needs to be chopped up into smaller pieces for service. This
283 may indicate a performance problem due to a bad setup of that raid/dm device,
284 but may also just be part of normal boundary conditions. dm is notably bad at
285 this and will clone lots of i/o.
289 For stacked devices, incoming i/o is remapped to device below it in the i/o
290 stack. The remap action details what exactly is being remapped to what.
295 .SH "OUTPUT DESCRIPTION AND FORMATTING"
297 The output from blkparse can be tailored for specific use -- in particular, to ease
298 parsing of output, and/or limit output fields to those the user wants to see. The
299 data for fields which can be output include:
302 Action, a (small) string (1 or 2 characters) -- see table below for more details
311 RWBS field, a (small) string (1-3 characters) -- see section below for more details
314 7-character string containing the major and minor numbers of
315 the event's device (separated by a comma).
321 Minor number of event's device.
324 Major number of event's device.
336 Display packet data \-\- series of hexadecimal values
345 Time stamp (nanoseconds)
351 Elapsed value in microseconds (\fI\-t\fR command line option)
354 Payload unsigned integer
357 Note that the user can optionally specify field display width, and optionally a
358 left-aligned specifier. These precede field specifiers, with a '%' character,
359 followed by the optional left-alignment specifier (\-) followed by the width (a
360 decimal number) and then the field.
362 Thus, to specify the command in a 12-character field that is left aligned:
367 .SH "ACTION IDENTIFIERS"
369 The following table shows the various actions which may be output:
372 IO was remapped to a different device
384 IO front merged with request on queue
390 IO inserted onto request queue
393 IO back merged with request on queue
399 IO handled by request queue code
405 Unplug due to timeout
414 .SH "RWBS DESCRIPTION"
416 This is a small string containing at least one character ('R' for read, 'W'
417 for write, or 'D' for block discard operation), and optionally either
418 a 'B' (for barrier operations) or 'S' (for synchronous operations).
423 The standard header (or initial fields displayed) include:
425 "%D %2c %8s %5T.%9t %5p %2a %3d"
430 Displays the event's device major/minor as: %3d,%\-3d.
433 CPU ID (2-character field).
439 5-character field for the seconds portion of the time stamp and a 9-character field for the nanoseconds in the time stamp.
442 5-character field for the process ID.
445 2-character field for one of the actions.
448 3-character field for the RWBS data.
450 Seeing this in action:
452 8,0 3 1 0.000000000 697 G W 223490 + 8 [kjournald]
454 The header is the data in this line up to the 223490 (starting block).
455 The default output for all event types includes this header.
459 .SH "DEFAULT OUTPUT PER ACTION"
461 \fBC \-\- complete\fR
463 If a payload is present, this is presented between
464 parenthesis following the header, followed by the error value.
466 If no payload is present, the sector and number of blocks are presented
467 (with an intervening plus (+) character). If the \fB\-t\fR option
468 was specified, then the elapsed time is presented. In either case,
469 it is followed by the error value for the completion.
476 \fBI \-\- inserted\fR
480 If a payload is present, the number of payload bytes
481 is output, followed by the payload in hexadecimal between parenthesis.
483 If no payload is present, the sector and number of blocks are presented
484 (with an intervening plus (+) character). If the \fB\-t\fR option was
485 specified, then the elapsed time is presented (in parenthesis). In
486 either case, it is followed by the command associated with the event
487 (surrounded by square brackets).
490 \fBF \-\- front merge\fR
492 \fBG \-\- get request\fR
494 \fBM \-\- back merge\fR
498 The starting sector and number of blocks is output
499 (with an intervening plus (+) character), followed by the command
500 associated with the event (surrounded by square brackets).
505 The command associated with the event (surrounded by
506 square brackets) is output.
511 \fBT \-\- unplug due to timer\fR
513 The command associated with the event
514 (surrounded by square brackets) is output, followed by the number of
515 requests outstanding.
520 The original starting sector followed by the new
521 sector (separated by a slash (/) is output, followed by the command
522 associated with the event (surrounded by square brackets).
527 Sector and length is output, along with the original
528 device and sector offset.
533 To trace the i/o on the device \fI/dev/hda\fB and parse the output to human
534 readable form, use the following command:
536 % blktrace \-d /dev/sda \-o \- | blkparse \-i \-
538 (see \fIblktrace\fR (8) for more information).
539 This same behaviour can be achieve with the convenience script \fIbtrace\fR.
544 has exactly the same effect as the previous command. See \fIbtrace\fR (8) for
547 To trace the i/o on a device and save the output for later processing with
548 \fIblkparse\fR, use \fIblktrace\fR like this:
550 % blktrace /dev/sda /dev/sdb
552 This will trace i/o on the devices \fI/dev/sda\fR and \fI/dev/sdb\fR and save
553 the recorded information in the files \fIsda\fR and \fIsdb\fR in the current
554 directory, for the two different devices, respectively. This trace
555 information can later be parsed by the \fIblkparse\fR utility:
559 which will output the previously recorded tracing information in human
560 readable form to stdout.
564 \fIblkparse\fR was written by Jens Axboe, Alan D. Brunelle and Nathan Scott. This
565 man page was created from the \fIblktrace\fR documentation by Bas Zoetekouw.
569 Report bugs to <linux\-btrace@vger.kernel.org>
572 Copyright \(co 2006 Jens Axboe, Alan D. Brunelle and Nathan Scott.
574 This is free software. You may redistribute copies of it under the terms of
575 the GNU General Public License <http://www.gnu.org/licenses/gpl.html>.
576 There is NO WARRANTY, to the extent permitted by law.
578 This manual page was created for Debian by Bas Zoetekouw. It was derived from
579 the documentation provided by the authors and it may be used, distributed and
580 modified under the terms of the GNU General Public License, version 2.
582 On Debian systems, the text of the GNU General Public License can be found in
583 /usr/share/common\-licenses/GPL\-2.
586 btrace (8), blktrace (8), verify_blkparse (1), blkrawverify (1), btt (1)