On 25/04/2018, Mohamad Gebai wrote: > It is very fast, but the downside would be usability. To move all logs > to LTTng tracepoints means that someone needs to create a tracing > session, make sure the lttng-session daemon is running, enable the right > events, start the tracer (this can be done in Ceph using the LTTng's C > API), think about what happens when a Ceph process crashes, what happens > when LTTng crashes, what happens if someone on the system inadvertently > stops the tracing system-wide, etc. Some of these issues can be > overcome, though. In summary: it's not the default use case of LTTng, > but there is a way if we wanted to go down that path. > I've heard of a suggestion that we could also use a mix of both: dout() > for lower log_levels and LTTng for for higher ones. This at least sounds like something we could handle fairly well in the default case by writing our systemd services to pull everything up and restart pieces on failure. -- Senior Software Engineer Red Hat Storage, Ann Arbor, MI, US IRC: Aemerson@OFTC, Actinic@Freenode 0x80F7544B90EDBFB9 E707 86BA 0C1B 62CC 152C 7C12 80F7 544B 90ED BFB9 -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
