linux-trace-users.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* stream based live time synchronized tracing?
@ 2022-08-15 13:28 Alexander Aring
  2022-08-15 15:02 ` Mathieu Desnoyers
  2022-08-16 22:05 ` Steven Rostedt
  0 siblings, 2 replies; 3+ messages in thread
From: Alexander Aring @ 2022-08-15 13:28 UTC (permalink / raw)
  To: linux-trace-users

Hi,

I am curious if there is any way of getting time synchronized traces
in a kind of stream based communication like pipes, sockets, etc. and
get high level event representation as ?libtraceevent? provides. I
would like to get all "merged" events from all machines provided by -A
parameters. I think it isn't required to get them in order, but the
timestamp should be synchronized.

I could probably build something like that with libtracecmd to have it
directly implemented in an application, but I am curious if there
exists any interest in having such a feature upstream? E.g. having
additional parameters like '--tsync-interval'.

A use-case would be live capturing of time synchronized events or what
I have in my mind to collect stats and doing kernel runtime
optimizations.

- Alex


^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: stream based live time synchronized tracing?
  2022-08-15 13:28 stream based live time synchronized tracing? Alexander Aring
@ 2022-08-15 15:02 ` Mathieu Desnoyers
  2022-08-16 22:05 ` Steven Rostedt
  1 sibling, 0 replies; 3+ messages in thread
From: Mathieu Desnoyers @ 2022-08-15 15:02 UTC (permalink / raw)
  To: Alexander Aring; +Cc: linux-trace-users

----- On Aug 15, 2022, at 9:28 AM, Alexander Aring aahringo@redhat.com wrote:

> Hi,
> 
> I am curious if there is any way of getting time synchronized traces
> in a kind of stream based communication like pipes, sockets, etc. and
> get high level event representation as ?libtraceevent? provides. I
> would like to get all "merged" events from all machines provided by -A
> parameters. I think it isn't required to get them in order, but the
> timestamp should be synchronized.
> 
> I could probably build something like that with libtracecmd to have it
> directly implemented in an application, but I am curious if there
> exists any interest in having such a feature upstream? E.g. having
> additional parameters like '--tsync-interval'.
> 
> A use-case would be live capturing of time synchronized events or what
> I have in my mind to collect stats and doing kernel runtime
> optimizations.

You may want to have a look at babeltrace2 [1]. It allows merging multiple
traces together as they are read. Using the "lttng-live" source, you can read
traces as they are produced.

Thanks,

Mathieu

[1] https://babeltrace.org


-- 
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: stream based live time synchronized tracing?
  2022-08-15 13:28 stream based live time synchronized tracing? Alexander Aring
  2022-08-15 15:02 ` Mathieu Desnoyers
@ 2022-08-16 22:05 ` Steven Rostedt
  1 sibling, 0 replies; 3+ messages in thread
From: Steven Rostedt @ 2022-08-16 22:05 UTC (permalink / raw)
  To: Alexander Aring; +Cc: linux-trace-users

On Mon, 15 Aug 2022 09:28:48 -0400
Alexander Aring <aahringo@redhat.com> wrote:

> Hi,
> 
> I am curious if there is any way of getting time synchronized traces
> in a kind of stream based communication like pipes, sockets, etc. and
> get high level event representation as ?libtraceevent? provides. I
> would like to get all "merged" events from all machines provided by -A
> parameters. I think it isn't required to get them in order, but the
> timestamp should be synchronized.

Are these for guests, or machines on a network?

> 
> I could probably build something like that with libtracecmd to have it
> directly implemented in an application, but I am curious if there
> exists any interest in having such a feature upstream? E.g. having
> additional parameters like '--tsync-interval'.
> 
> A use-case would be live capturing of time synchronized events or what
> I have in my mind to collect stats and doing kernel runtime
> optimizations.

So basically not save to a file? Just record the events live and interleave
them for different machines?

I'm not exactly sure what you are asking for. Can you please elaborate some
more?

Thanks,

-- Steve


^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2022-08-16 22:05 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-08-15 13:28 stream based live time synchronized tracing? Alexander Aring
2022-08-15 15:02 ` Mathieu Desnoyers
2022-08-16 22:05 ` Steven Rostedt

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).