[lttng-dev] [LTTNG] Empty traces output file

[Jonathan Rajotte-Julien]

On Fri, Feb 09, 2018 at 11:07:47PM -0500, Shieryn Tjandra wrote:
> Good Evening,
> 
> Hi There,
> 
> I have inquiries regarding LTTNG, I followed the tutorial in this link :
> 
> http://lttng.org/docs/v2.10/#doc-tracing-your-own-user-application

Does it work with the basic application described in this section ? Without
thread and all?

> 
> However, after executing the c application, when I checked whether the
> process was in the userspace event (UST), the process was not there. Next,
> I tried to record the  traces, as I thought, my traces output file does not
> contain any traces. I also have make sure that liblttng-ust has been
> installed. I really appreciate if you can help me to overcome this problem,
> if you need additional file, let me know.
> 
> I am using centos 7 by the way.
> 
> Here is my command line
> [root at localhost tracepractice]# gcc -c -I. app_tp.c
> [root at localhost tracepractice]# gcc -c app.c -lpthread
> [root at localhost tracepractice]# gcc -o try app.o app_tp.o -lpthread
> -llttng-ust -ldl
> [root at localhost tracepractice]# ./try and beyond
> Main thread:     : pid = 17231 lwp = 17231 priority = 0
> Thread 4:     : pid = 17231 lwp = 17234 priority = 0
> Thread 2:     : pid = 17231 lwp = 17236 priority = 8
> Thread 1:     : pid = 17231 lwp = 17237 priority = 12
> Thread 1: Hello! I am thread 4 and I more important than 1, 2 and 3.     My
> count is now 0
> Thread 3:     : pid = 17231 lwp = 17235 priority = 4
> Thread 4: Hello! I am thread 1 and I not very important.         My count
> is now 1
> Thread 2: Hello! I am thread 3 and I more important than 1 and 2.     My
> count is now 0
> Thread 3: Hello! I am thread 2 and I more important than 1.         My
> count is now 1
> Thread 2: Hello! I am thread 3 and I more important than 1 and 2.     My
> count is now 262144
> Thread 3: Hello! I am thread 2 and I more important than 1.         My
> count is now 262145
> sig_ala [cound down=1]
> sig_ala [cound down=0]
> [root at localhost tracepractice]# lttng-sessiond --daemonize
> [root at localhost tracepractice]# lttng list --userspace
> UST events:
> -------------
> None

Is your application running at that point?

If not, "lttng list --userspace" will only list tracepoints for ongoing
application.

> 
> [root at localhost tracepractice]# lttng create mytrace
> Session mytrace created.
> Traces will be written in /root/lttng-traces/mytrace-20180209-013417
> [root at localhost tracepractice]# lttng enable-event --userspace
> app_tp:first_tracepoint
> UST event app_tp:first_tracepoint created in channel channel0
> [root at localhost tracepractice]# lttng start
> Tracing started for session mytrace
> [root at localhost tracepractice]# ./try

Could you launch your application with LTTNG_UST_DEBUG=y and provide us the
resulting output?

e.g:
LTTNG_UST_DEBUG=y ./try

Does your application uses fork()?

Could you provide the source code of your test application (github?).

> Main thread:     : pid = 17301 lwp = 17301 priority = 0
> Thread 4:     : pid = 17301 lwp = 17304 priority = 0
> Thread 2:     : pid = 17301 lwp = 17306 priority = 8
> Thread 1:     : pid = 17301 lwp = 17307 priority = 12
> Thread 1: Hello! I am thread 4 and I more important than 1, 2 and 3.     My
> count is now 0
> Thread 4: Hello! I am thread 1 and I not very important.         My count
> is now 1
> Thread 3:     : pid = 17301 lwp = 17305 priority = 4
> Thread 2: Hello! I am thread 3 and I more important than 1 and 2.     My
> count is now 0
> Thread 3: Hello! I am thread 2 and I more important than 1.         My
> count is now 1
> Thread 2: Hello! I am thread 3 and I more important than 1 and 2.     My
> count is now 262144
> Thread 3: Hello! I am thread 2 and I more important than 1.         My
> count is now 262145
> sig_ala [cound down=1]
> Thread 2: Hello! I am thread 3 and I more important than 1 and 2.     My
> count is now 524288
> Thread 3: Hello! I am thread 2 and I more important than 1.         My
> count is now 524289
> sig_ala [cound down=0]
> [root at localhost tracepractice]# lttng stop
> Waiting for data availability
> Tracing stopped for session mytrace
> [root at localhost tracepractice]# babeltrace ~/lttng-traces/mytrace*
> [root at localhost tracepractice]# ls ~/lttng-traces/mytrace* -l
> total 0
> drwxrwx---. 3 root root 17 Feb  9 01:35 ust

Actual trace files are normally deeper in the trace directory.

e.g
$ tree -h lttng-traces/auto-20180207-150736/

lttng-traces/auto-20180207-150736/
└── [4.0K]  ust
    └── [4.0K]  uid
        └── [4.0K]  1000
            └── [4.0K]  64-bit
                ├── [296K]  channel0_0
                ├── [576K]  channel0_1
                ├── [300K]  channel0_2
                ├── [400K]  channel0_3
                ├── [4.0K]  index
                │   ├── [ 232]  channel0_0.idx
                │   ├── [ 376]  channel0_1.idx
                │   ├── [ 232]  channel0_2.idx
                │   └── [ 304]  channel0_3.idx
                └── [ 16K]  metadata

5 directories, 9 files

Cheers

-- 
Jonathan Rajotte-Julien
EfficiOS