Re: tracepoint_enabled returns true for disabled provider, event

Re: tracepoint_enabled returns true for disabled provider, event

[Jonathan Rajotte-Julien]

Hi,

In the end this was a problem involving automatic session creation
(configuration) behind the back of Anders.

A "lttng list" helped figure out the problem.

lttng behave as expected.

Cheers

On Sat, Sep 07, 2019 at 02:42:22PM +0200, Anders Johansson wrote:
> Hi
> 
> I troubleshooting a huge application where i experience the problem that
> even if tracepoint_enabled is used i get the huge overhead of formatting
> the output even if then the message is not sent to the lttng buffers.
> 
> The tracepoint loglevel is set to TRACE_INFO using the TRACEPOINT_LOGLEVEL
> macro. I managed to rebuild the application setting the level to
> TRACE_DEBUG instead and then i could print continuously that the
> tracepoint_enabled returned false. If i flip it back to TRACE_INFO
> tracepoint_enabled always returns true, hence do lots of string building
> but then no trace is actually sent off to lttng (which is correct since i
> haven't enabled the tracepoint). Even if i explicitly do disable event on
> the provider and tracepoint the TRACE_INFO tracepoints returns true in
> tracepoint_enabled
> 
> Is this by design that tracepoint_enabled always returns true for
> TRACE_INFO and lower values or is this configurable in any way?
> 
> I am using this inside a docker image running Wind River Linux
> bash-4.3# lttng --version
> lttng (LTTng Trace Control) 2.10.6 - KeKriek - v2.10.6-11-g8fcbf96
> bash-4.3#
> 
> Any advice are welcome!
> //Anders

> _______________________________________________
> lttng-dev mailing list
> lttng-dev@lists.lttng.org
> https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev


-- 
Jonathan Rajotte-Julien
EfficiOS

tracepoint_enabled returns true for disabled provider, event

[Anders Johansson]

[-- Attachment #1.1: Type: text/plain, Size: 1136 bytes --]

Hi

I troubleshooting a huge application where i experience the problem that
even if tracepoint_enabled is used i get the huge overhead of formatting
the output even if then the message is not sent to the lttng buffers.

The tracepoint loglevel is set to TRACE_INFO using the TRACEPOINT_LOGLEVEL
macro. I managed to rebuild the application setting the level to
TRACE_DEBUG instead and then i could print continuously that the
tracepoint_enabled returned false. If i flip it back to TRACE_INFO
tracepoint_enabled always returns true, hence do lots of string building
but then no trace is actually sent off to lttng (which is correct since i
haven't enabled the tracepoint). Even if i explicitly do disable event on
the provider and tracepoint the TRACE_INFO tracepoints returns true in
tracepoint_enabled

Is this by design that tracepoint_enabled always returns true for
TRACE_INFO and lower values or is this configurable in any way?

I am using this inside a docker image running Wind River Linux
bash-4.3# lttng --version
lttng (LTTng Trace Control) 2.10.6 - KeKriek - v2.10.6-11-g8fcbf96
bash-4.3#

Any advice are welcome!
//Anders

[-- Attachment #1.2: Type: text/html, Size: 1348 bytes --]

[-- Attachment #2: Type: text/plain, Size: 156 bytes --]

_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev