I seem to have gotten my lttng into a state where kernel tracing does not work.
I have tried uninstalling and reinstalling packages with no luck and also
building from source.
Any suggestions welcome.
My distro is Ubuntu 16.04 but the kernel is 4.10.0-20-generic.
I have installed 2.9 using
apt i
Thanks, the new packages installed fine.
My original problem I solved by killing lttng-sessiond and restarting it...
-- Tom
-Original Message-
From: Jonathan Rajotte-Julien [mailto:jonathan.rajotte-jul...@efficios.com]
Sent: Wednesday, May 31, 2017 1:30 PM
To: Deneau, Tom
Cc: lttng
Using, lttng 2.9.4
Doing combined kernel and usermode tracing and then looking at the combined
trace with babeltrace...
I have noticed there is often a period at the beginning of the combined trace
which is all kernel events and similarly a period at the end of the combined
trace which is all u
I am using lttng 2.9.4.
I have a server database daemon with many threads and I would like to track
events on a smaller subset of those threads.
I noticed that on the kernel channel I can use lttng track --pid=123,124,125
and this will trace kernel events for just those TIDS (in spite of the
do
e-
> From: Mathieu Desnoyers [mailto:mathieu.desnoy...@efficios.com]
> Sent: Thursday, June 01, 2017 1:25 PM
> To: Deneau, Tom
> Cc: lttng-dev
> Subject: Re: [lttng-dev] lttng-track on userspace channel
>
> - On May 31, 2017, at 7:28 PM, Deneau, Tom tom.den...@amd.
I am using the babeltrace python3 bindings on a trace collected with
event-loss-mode
set to discard. The tracecollection is a combination of a kernel channel and a
usermode
channel. How can I tell whether there are any lost events in this trace?
-- Tom Deneau
__
I am trying to use lttng to simultaneously trace some kernel and user events.
The kernel events are a few syscalls (sendto, futex).
The user events are the pthread ones in lttng-ust-libc-wrapper.
Looking for recommended lttng configurations to:
* avoid dropping any events
* avoid too much ov
Is there a write-up on lttng timestamp precision?
I have noticed on a 4.10 kernel with a mix of kernel and usermode events
that the minimum time between events on a single software thread seems to be
about 400 ns.
I noticed on this same system that clock_getres(CLOCK_MONOTONIC) reports 1 ns
res