1 .TH "LTTNG" "1" "December 3rd, 2012" "" ""
4 lttng \(em LTTng 2.x tracer control command line tool
10 lttng [OPTIONS] <COMMAND>
15 The LTTng project aims at providing highly efficient tracing tools for Linux.
16 It's tracers help tracking down performance issues and debugging problems
17 involving multiple concurrent processes and threads. Tracing across multiple
18 systems is also possible.
20 The \fBlttng\fP command line tool from the lttng-tools package is used to control
21 both kernel and user-space tracing. Every interactions with the tracer should
22 be done by this tool or by the liblttng-ctl provided with the lttng-tools
25 LTTng uses a session daemon (lttng-sessiond(8)), acting as a tracing registry,
26 which allows you to interact with multiple tracers (kernel and user-space)
27 inside the same container, a tracing session. Traces can be gathered from the
28 kernel and/or instrumented applications (lttng-ust(3)). Aggregating and reading
29 those traces is done using the babeltrace(1) text viewer.
31 We introduce the notion of \fBtracing domains\fP which is essentially a type of
32 tracer (kernel or user space for now). In the future, we could see a third
33 tracer being for instance an hypervisor. For some commands, you'll need to
34 specify on which domain the command applies (-u or -k). For instance, enabling
35 a kernel event, you must specify the kernel domain to the command so we know
36 for which tracer this event is for.
38 In order to trace the kernel, the session daemon needs to be running as root.
39 LTTng provides the use of a \fBtracing group\fP (default: tracing). Whomever is
40 in that group can interact with the root session daemon and thus trace the
41 kernel. Session daemons can co-exist meaning that you can have a session daemon
42 running as Alice that can be used to trace her applications along side with a
43 root daemon or even a Bob daemon. We highly recommend to start the session
44 daemon at boot time for stable and long term tracing.
46 Every user-space applications instrumented with lttng-ust(3), will
47 automatically register to the session daemon. This feature gives you the
48 ability to list available traceable applications and tracepoints on a per user
49 basis. (See \fBlist\fP command).
53 This program follow the usual GNU command line syntax with long options starting with
54 two dashes. Below is a summary of the available options.
59 Show summary of possible options and commands.
61 .BR "\-v, \-\-verbose"
63 Three levels of verbosity are available which are triggered by putting additional v to
64 the option (\-vv or \-vvv)
67 Suppress all messages (even errors).
69 .BR "\-g, \-\-group NAME"
70 Set unix tracing group name. (default: tracing)
72 .BR "\-n, \-\-no-sessiond"
73 Don't automatically spawn a session daemon.
75 .BR "\-\-sessiond\-path PATH"
76 Set session daemon full binary path.
78 .BR "\-\-list\-options"
79 Simple listing of lttng options.
81 .BR "\-\-list\-commands"
82 Simple listing of lttng commands.
88 Add context to event(s) and/or channel(s).
90 A context is basically extra information appended to a channel. For instance,
91 you could ask the tracer to add the PID information for all events in a
92 channel. You can also add performance monitoring unit counters (perf PMU) using
95 For example, this command will add the context information 'prio' and two perf
96 counters (hardware branch misses and cache misses), to all events in the trace
99 # lttng add-context \-k \-t prio \-t perf:branch-misses \-t perf:cache-misses
101 Please take a look at the help (\-h/\-\-help) for a detailed list of available
104 If no channel is given (\-c), the context is added to all channels that were
105 already enabled. If the session has no channel, a default channel is created.
106 Otherwise the context will be added only to the given channel (\-c).
108 If \fB\-s, \-\-session\fP is omitted, the session name is taken from the .lttngrc
116 Show summary of possible options and commands.
117 \-s, \-\-session NAME
118 Apply on session name.
119 \-c, \-\-channel NAME
120 Apply on channel name.
122 Apply for the kernel tracer
124 Apply for the user-space tracer
126 Context type. You can repeat this option on the command line. Please
127 use "lttng add-context \-h" to list all available types.
132 .IP "\fBcalibrate\fP"
134 Quantify LTTng overhead
136 The LTTng calibrate command can be used to find out the combined average
137 overhead of the LTTng tracer and the instrumentation mechanisms used. This
138 overhead can be calibrated in terms of time or using any of the PMU performance
139 counter available on the system.
141 For now, the only calibration implemented is that of the kernel function
142 instrumentation (kretprobes).
144 * Calibrate kernel function instrumentation
146 Let's use an example to show this calibration. We use an i7 processor with 4
147 general-purpose PMU registers. This information is available by issuing dmesg,
148 looking for "generic registers".
150 This sequence of commands will gather a trace executing a kretprobe hooked on
151 an empty function, gathering PMU counters LLC (Last Level Cache) misses
152 information (see lttng add-context \-\-help to see the list of available PMU
155 # lttng create calibrate-function
156 # lttng enable-event calibrate \-\-kernel \-\-function lttng_calibrate_kretprobe
157 # lttng add-context \-\-kernel \-t perf:LLC-load-misses \-t perf:LLC-store-misses \\
158 \-t perf:LLC-prefetch-misses
160 # for a in $(seq 1 10); do \\
161 lttng calibrate \-\-kernel \-\-function;
164 # babeltrace $(ls \-1drt ~/lttng-traces/calibrate-function-* | tail \-n 1)
166 The output from babeltrace can be saved to a text file and opened in a
167 spreadsheet (e.g. oocalc) to focus on the per-PMU counter delta between
168 consecutive "calibrate_entry" and "calibrate_return" events. Note that these
169 counters are per-CPU, so scheduling events would need to be present to account
170 for migration between CPU. Therefore, for calibration purposes, only events
171 staying on the same CPU must be considered.
173 The average result, for the i7, on 10 samples:
176 perf_LLC_load_misses: 5.0 0.577
177 perf_LLC_store_misses: 1.6 0.516
178 perf_LLC_prefetch_misses: 9.0 14.742
180 As we can notice, the load and store misses are relatively stable across runs
181 (their standard deviation is relatively low) compared to the prefetch misses.
182 We can conclude from this information that LLC load and store misses can be
183 accounted for quite precisely, but prefetches within a function seems to behave
184 too erratically (not much causality link between the code executed and the CPU
185 prefetch activity) to be accounted for.
192 Show summary of possible options and commands.
194 Apply for the kernel tracer
196 Apply for the user-space tracer
198 Dynamic function entry/return probe (default)
203 .IP "\fBcreate\fP [NAME] [OPTIONS]
205 Create tracing session.
207 A tracing session contains channel(s) which contains event(s). It is domain
208 agnostic meaning that you can enable channels and events for either the
209 user-space tracer and/or the kernel tracer. It acts like a container
210 aggregating multiple tracing sources.
212 On creation, a \fB.lttngrc\fP file is created in your $HOME directory
213 containing the current session name. If NAME is omitted, a session name is
214 automatically created having this form: 'auto-yyyymmdd-hhmmss'.
216 If no \fB\-o, \-\-output\fP is specified, the traces will be written in
219 The $HOME environment variable can be overridden by defining the environment
220 variable LTTNG_HOME. This is useful when the user running the commands has
221 a non-writeable home directory.
228 Show summary of possible options and commands.
230 Simple listing of options
232 Specify output path for traces
234 Using these options, each API call can be controlled individually. For
235 instance, \-C does not enable the consumer automatically. You'll need the \-e
239 Set URL for the consumer output destination. It is persistent for the
240 session lifetime. Redo the command to change it. This will set both
241 data and control URL for network.
242 \-C, \-\-ctrl-url=URL
243 Set control path URL. (Must use -D also)
244 \-D, \-\-data-url=URL
245 Set data path URL. (Must use -C also)
249 proto://[HOST|IP][:PORT1[:PORT2]][/TRACE_PATH]
251 Supported protocols are (proto):
253 Local filesystem full path.
256 This will use the default network transport layer which is TCP for both
257 control (PORT1) and data port (PORT2). The default ports are
258 respectively 5342 and 5343. Note that net[6]:// is not yet supported.
261 Can only be used with -C and -D together
263 NOTE: IPv6 address MUST be enclosed in brackets '[]' (rfc2732)
267 # lttng create -U net://192.168.1.42
268 Uses TCP and default ports for the given destination.
270 # lttng create -U net6://[fe80::f66d:4ff:fe53:d220]
271 Uses TCP, default ports and IPv6.
273 # lttng create s1 -U net://myhost.com:3229
274 Create session s1 and set its consumer to myhost.com on port 3229 for control.
279 .IP "\fBdestroy\fP [OPTIONS] [NAME]"
281 Teardown tracing session
283 Free memory on the session daemon and tracer side. It's gone!
285 If NAME is omitted, the session name is taken from the .lttngrc file.
292 Show summary of possible options and commands.
296 Simple listing of options
301 .IP "\fBenable-channel\fP NAME[,NAME2,...] [-k|-u] [OPTIONS]"
303 Enable tracing channel
305 To enable an event, you must enable both the event and the channel that
308 If \fB\-s, \-\-session\fP is omitted, the session name is taken from the .lttngrc
311 It is important to note that if a certain type of buffers is used, the session
312 will be set with that type and all other subsequent channel needs to have the
315 For the kernel tracer (\-k), once the session is started at least once, it's
316 not possible anymore to enable new channel for that session.
325 Simple listing of options
326 \-s, \-\-session NAME
327 Apply on session name
329 Apply to the kernel tracer
331 Apply to the user-space tracer
334 Discard event when subbuffers are full (default)
336 Flight recorder mode : overwrites events when subbuffers are full
338 Subbuffer size in bytes {+k,+M,+G}
339 (default UST uid: 131072, UST pid: 4096, kernel: 262144, metadata: 4096)
340 Rounded up to the next power of 2.
342 The minimum subbuffer size, for each tracer, is the max value between
343 the default above and the system page size. You can issue this command
344 to get the current page size on your system: \fB$ getconf PAGE_SIZE\fP
346 Number of subbuffers. (default UST uid: 4, UST pid: 4, kernel: 4, metadata: 2)
347 Rounded up to the next power of 2.
348 \-\-switch-timer USEC
349 Switch subbuffer timer interval in µsec.
350 (default UST uid: 0, UST pid: 0, kernel: 0, metadata: 0)
352 Read timer interval in µsec.
353 (default UST uid: 0, UST pid: 0, kernel: 200000, metadata: 0)
355 Channel output type. Possible values: mmap, splice
356 (default UST uid: mmap, UST pid: mmap, kernel: splice, metadata: mmap)
358 Use per UID buffer (\-u only). Buffers are shared between applications
359 that have the same UID.
361 Use per PID buffer (\-u only). Each application has its own buffers.
363 Use shared buffer for the whole system (\-k only)
364 \-C, \-\-tracefile-size SIZE
365 Maximum size of each tracefile within a stream (in bytes).
366 0 means unlimited. (default: 0)
367 \-W, \-\-tracefile-count COUNT
368 Used in conjunction with \-C option, this will limit the number
369 of files created to the specified count. 0 means unlimited. (default: 0)
373 $ lttng enable-channel -C 4096 -W 32 chan1
374 For each stream, the maximum size of a trace file will be 4096 bytes divided
375 over a \fBmaximum\fP of 32 different files. The file count is appended after
376 the stream number as seen in the following example. The last trace file is
377 smaller than 4096 since it was not completely filled.
379 ~/lttng-traces/[...]/chan1_0_0 (4096)
380 ~/lttng-traces/[...]/chan1_0_1 (4096)
381 ~/lttng-traces/[...]/chan1_0_2 (3245)
382 ~/lttng-traces/[...]/chan1_1_0 (4096)
385 $ lttng enable-channel -C 4096
386 This will create trace files of 4096 bytes and will create new ones as long as
387 there is data available.
392 .IP "\fBenable-event\fP NAME[,NAME2,...] [-k|-u] [OPTIONS]"
396 A tracing event is always assigned to a channel. If \fB\-c, \-\-channel\fP is
397 omitted, a default channel named '\fBchannel0\fP' is created and the event is
398 added to it. For the user-space tracer, using \fB\-a, \-\-all\fP is the same as
399 using the wildcard "*".
401 If \fB\-s, \-\-session\fP is omitted, the session name is taken from the .lttngrc
409 Show summary of possible options and commands.
411 Simple listing of options
412 \-s, \-\-session NAME
413 Apply on session name
414 \-c, \-\-channel NAME
415 Apply on channel name
417 Enable all tracepoints and syscalls. This actually enable a single
420 Apply for the kernel tracer
422 Apply for the user-space tracer
425 Tracepoint event (default)
426 - userspace tracer supports wildcards at end of string. Don't forget to
427 quote to deal with bash expansion.
432 Tracepoint loglevel range from 0 to loglevel. Listed in the help (\-h).
433 \-\-loglevel-only NAME
434 Tracepoint loglevel (only this loglevel).
436 The loglevel or loglevel-only options should be combined with a
437 tracepoint name or tracepoint wildcard.
438 \-\-probe [addr | symbol | symbol+offset]
439 Dynamic probe. Addr and offset can be octal (0NNN...), decimal (NNN...)
440 or hexadecimal (0xNNN...)
441 \-\-function [addr | symbol | symbol+offset]
442 Dynamic function entry/return probe. Addr and offset can be octal
443 (0NNN...), decimal (NNN...) or hexadecimal (0xNNN...)
445 System call event. Enabling syscalls tracing (kernel tracer), you will
446 not be able to disable them with disable-event. This is a known
447 limitation. You can disable the entire channel to do the trick.
449 \-\-filter 'expression'
450 Set a filter on a newly enabled event. Filter expression on event
451 fields and context. Event recording depends on evaluation. Only
452 specify on first activation of a given event within a session.
453 Filter only allowed when enabling events within a session before
454 tracing is started. If the filter fails to link with the event
455 within the traced domain, the event will be discarded.
456 Currently, filter is only implemented for the user-space tracer.
460 'intfield > 500 && intfield < 503'
461 '(stringfield == "test" || intfield != 10) && intfield > 33'
462 'doublefield > 1.1 && intfield < 5.3'
464 Wildcards are allowed at the end of strings:
466 In string literals, the escape character is a '\\'. Use '\\*' for
467 the '*' character, and '\\\\' for the '\\' character. Wildcard
468 match any sequence of characters, including an empty sub-string
469 (match 0 or more characters).
471 Context information can be used for filtering. The examples
472 below show usage of context filtering on process name (with a
473 wildcard), process ID range, and unique thread ID for filtering.
474 The process and thread ID of running applications can be found
475 under columns "PID" and "LWP" of the "ps -eLf" command.
477 '$ctx.procname == "demo*"'
478 '$ctx.vpid >= 4433 && $ctx.vpid < 4455'
482 .IP "\fBdisable-channel\fP NAME[,NAME2,...] [\-k|\-u] [OPTIONS]"
484 Disable tracing channel
486 Disabling a channel makes all event(s) in that channel to stop tracing. You can
487 enable it back by calling \fBlttng enable-channel NAME\fP again.
489 If \fB\-s, \-\-session\fP is omitted, the session name is taken from the .lttngrc
497 Show summary of possible options and commands.
499 Simple listing of options
500 \-s, \-\-session NAME
501 Apply on session name
503 Apply for the kernel tracer
505 Apply for the user-space tracer
508 .IP "\fBdisable-event\fP NAME[,NAME2,...] [\-k|\-u] [OPTIONS]"
510 Disable tracing event
512 The event, once disabled, can be re-enabled by calling \fBlttng enable-event
515 If \fB\-s, \-\-session\fP is omitted, the session name is taken from the .lttngrc
523 Show summary of possible options and commands.
525 Simple listing of options
526 \-s, \-\-session NAME
527 Apply on session name
529 Disable all events. This does NOT disable "*" but rather
530 every known events of the session.
532 Apply for the kernel tracer
534 Apply for the user-space tracer
537 .IP "\fBlist\fP [\-k|\-u] [SESSION [SESSION_OPTIONS]]"
539 List tracing session information.
541 With no arguments, it will list available tracing session(s).
543 With the session name, it will display the details of the session including
544 the trace file path, the associated channels and their state (activated
545 and deactivated), the activated events and more.
547 With \-k alone, it will list all available kernel events (except the system
549 With \-u alone, it will list all available user-space events from registered
550 applications. Here is an example of 'lttng list \-u':
552 PID: 7448 - Name: /tmp/lttng-ust/tests/hello/.libs/lt-hello
553 ust_tests_hello:tptest_sighandler (type: tracepoint)
554 ust_tests_hello:tptest (type: tracepoint)
556 You can now enable any event listed by using the name :
557 \fBust_tests_hello:tptest\fP.
564 Show summary of possible options and commands.
566 Simple listing of options
570 Select user-space domain.
574 \-c, \-\-channel NAME
575 List details of a channel
577 List available domain(s)
580 .IP "\fBset-session\fP NAME"
582 Set current session name
584 Will change the session name in the .lttngrc file.
591 Show summary of possible options and commands.
593 Simple listing of options
598 .IP "\fBstart\fP [NAME] [OPTIONS]"
602 It will start tracing for all tracers for a specific tracing session.
604 If NAME is omitted, the session name is taken from the .lttngrc file.
611 Show summary of possible options and commands.
613 Simple listing of options
618 .IP "\fBstop\fP [NAME] [OPTIONS]"
622 It will stop tracing for all tracers for a specific tracing session. Before
623 returning, the command checks for data availability meaning that it will wait
624 until the trace is readable for the session. Use \-\-no-wait to avoid this
627 If NAME is omitted, the session name is taken from the .lttngrc file.
634 Show summary of possible options and commands.
636 Simple listing of options
638 Don't wait for data availability.
645 Show version information
652 Show summary of possible options and commands.
654 Simple listing of options
659 .IP "\fBview\fP [SESSION_NAME] [OPTIONS]"
661 View traces of a tracing session
663 By default, the babeltrace viewer will be used for text viewing.
665 If SESSION_NAME is omitted, the session name is taken from the .lttngrc file.
675 Simple listing of options
676 \-t, \-\-trace-path PATH
677 Trace directory path for the viewer
679 Specify viewer and/or options to use
680 This will completely override the default viewers so
681 please make sure to specify the full command. The trace
682 directory path of the session will be appended at the end
687 On success 0 is returned and a positive value on error. Value of 1 means a command
688 error, 2 an undefined command, 3 a fatal error and 4 a command warning meaning that
689 something went wrong during the command.
691 Any other value above 10, please refer to
692 .BR <lttng/lttng-error.h>
693 for a detailed list or use lttng_strerror() to get a human readable string of
697 .SH "ENVIRONMENT VARIABLES"
700 Note that all command line options override environment variables.
704 .IP "LTTNG_SESSIOND_PATH"
705 Allows one to specify the full session daemon binary path to lttng command line
706 tool. You can also use \-\-sessiond-path option having the same effect.
710 .BR lttng-sessiond(8),
712 .BR lttng-health-check(3)
715 If you encounter any issues or usability problem, please report it on our
716 mailing list <lttng-dev@lists.lttng.org> to help improve this project or
717 at https://bugs.lttng.org which is a bugtracker.
721 lttng is distributed under the GNU General Public License version 2. See the file
724 A Web site is available at http://lttng.org for more information on the LTTng
727 You can also find our git tree at http://git.lttng.org.
729 Mailing lists for support and development: <lttng-dev@lists.lttng.org>.
731 You can find us on IRC server irc.oftc.net (OFTC) in #lttng.
736 Thanks to Yannick Brosseau without whom this project would never have been so
737 lean and mean! Also thanks to the Ericsson teams working on tracing which
738 helped us greatly with detailed bug reports and unusual test cases.
740 Thanks to our beloved packager Alexandre Montplaisir-Goncalves (Ubuntu and PPA
741 maintainer) and Jon Bernard for our Debian packages.
743 Special thanks to Michel Dagenais and the DORSAL laboratory at Polytechnique de
744 Montreal for the LTTng journey.
749 lttng-tools was originally written by Mathieu Desnoyers, Julien Desfossez and
750 David Goulet. More people have since contributed to it. It is currently
751 maintained by David Goulet <dgoulet@efficios.com>.