blog post: Monitoring real-time latencies

blog post: Monitoring real-time latencies

[Julien Desfossez]

Hi,

Here is a blog post related to detecting and understanding high
interrupt-processing latencies on real-time systems. It is based on a
new project called latency_tracker that hooks on the existing kernel
tracepoints and executes actions when high latency events occur.

https://lttng.org/blog/2016/01/06/monitoring-realtime-latencies/

It is a work in progress, so if you are interested and/or have any
comments, please let me know.

Thanks,

Julien

Re: blog post: Monitoring real-time latencies

[Alexander Hirsch]

Julien Desfossez <jdesfossez <at> efficios.com> writes:

> 
> Hi,
> 
> Here is a blog post related to detecting and understanding high
> interrupt-processing latencies on real-time systems. It is based on a
> new project called latency_tracker that hooks on the existing kernel
> tracepoints and executes actions when high latency events occur.
> 
> https://lttng.org/blog/2016/01/06/monitoring-realtime-latencies/
> 
> It is a work in progress, so if you are interested and/or have any
> comments, please let me know.
> 
> Thanks,
> 
> Julien
> 

Building lttng-modules 2.7.1 against linux-rt-4.1.15-rt17 (the most recent
releases respectively, except for linux-rt-4.4-rc*) fails because of the
patch introducing the sched_waking tracepoint to linux-rt
(sched-introduce-the-27trace_sched_waking-27-tracepoint.patch), which also
updates the sched_wakeup and sched_wakeup_template function prototypes to
linux-4.3.
The check for LINUX_VERSION_CODE >= KERNEL_VERSION(4,3,0) in the
instrumentation by lttng-modules
(instrumentation/events/lttng-module/sched.h) fails of course.

Regards,
Alexander Hirsch