babeltrace of flight recorder trace undergoes time warp?
Assume a flight recorder trace is generated like this:
$ lttng create limkern Session limkern created Traces will be written in /home/daniel/lttng-traces/limkern-20140123-085831 $ lttng enable-channel channel0 -k --tracefile-size 262144 --tracefile-count 1 Kernel channel channel0 enabled for session limkern $ lttng enable-event -k --all -c channel0 All kernel events are enabled in channel channel0 $ lttng start (wait for a while) $ lttng stop
Now when this is babeltraced:
$ babeltrace /home/daniel/lttng-traces/limkern-20140123-085831 -n all -f all -v -w /home/daniel/lttng-traces/limkern_nafav [verbose] Verbose mode active [...] [warning] Tracer discarded 378942 events between [09:00:44.643712446] and [09:00:44.643710896] in trace UUID ce[...]22, at path: "/home/daniel/lttng-traces/limkern-20140123-085831/kernel", within stream id 0, at relative path: "channel_0_0_0". You should consider recording a new trace with larger buffers or with fewer events enabled. [verbose] finished converting. Output written to: /home/daniel/lttng-traces/limkern_nafav
Note how the lost events are said to lie in a nonsensical time span where the end (44.64371*0896*) is before the start (44.64371*2446*).
Using babeltrace 1.1.1-31-513413e.
Related to bug #589?
Updated by Mathieu Desnoyers over 6 years ago
Can you confirm that this commit fixes the issue ?
commit 2654fe9bab8f0eaeb17264ef7abadfd14e245b23 Author: Mathieu Desnoyers <email@example.com> Date: Tue Feb 25 17:36:11 2014 -0500 Fix: events discarded timing inaccuracy Introduce babeltrace_ctf_console_output too, to ensure we only print discarded events when a text output is used. Fixes #589 Signed-off-by: Mathieu Desnoyers <firstname.lastname@example.org>