Document behavior of per-uid UST buffers with respect to asynchronously killed traced applications
We should document a known design limitation of the per-uid UST buffers:
When using lttng-ust with per-uid buffers, there is a known design limitation
regarding process terminated by non-handled kill signals: if this occurs while
the buffer is being written to (between reserve and commit), it will cause the
rest of the per-cpu buffer to be unreadable. This applies to all events
recorded by all applications of the same user id for that particular CPU.
Recovering from this involves destroying and re-creating the tracing session.
If this kind of scenario is likely for you, you might want to consider using per-
pid buffers (lttng enable-channel -u --buffers-pid), which do not suffer from
this design limitation. The downside of per-pid buffers is that it allocates
more memory on your system for buffering, and adds extra overhead when
used with processes that have a short life-time.
No data to display