[lttng-dev] Regarding lttng ust trace-point usage

[Jérémie Galarneau]

On 14 August 2018 at 08:06, jagadeesh konakalla <kjagadeesh27 at gmail.com> wrote:
> Hi Mathieu,
>
> This is Jagadeesh. I am user of lttng tools.
>
> I have one query regarding the lttng ust usage.
>
> I have written my linux application which has trace point API's to trace the
> user application.
> My requirement is to do both kernel domain trace and user domain trace for
> my application.
>
> When i start the lttng and then run my application, i do not see any issues.
> I mean to say that i get all the user trace events and kernel events.
>
> If i start the application, then i configure and start the lttng, i see some
> user events are not captured at the start (may be dropped). I have observed
> that there is a delay in the first kernel event occurrence to first user
> event occurrence. I have not seen this behavior if i start the lttng and run
> my application.

Hi,

It is expected for the kernel tracer to start tracing before the user space
tracer during the execution of the start command. Starting the tracing
across domains is not done in a single operation.

LTTng guarantees that all events that occur after the 'lttng start' command
returns will be captured (provided that the buffers are not full).

Jérémie

>
> Here is my question : Is this the expected behavior ?
> or
> Are there any settings that can help me to resolve this issue ?
>
> Version of lttng used :
> lttng-tools (2.9.4)
> lttng-modules (2.9.1)
> lttng-libust (2.9.0 )
>
> I appreciate your help. Please let me know if you need more information.
>
> With regards,
> Jagadeesh K.
>
> _______________________________________________
> lttng-dev mailing list
> lttng-dev at lists.lttng.org
> https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
>



-- 
Jérémie Galarneau
EfficiOS Inc.
http://www.efficios.com