[LITMUS^RT] high miss ratio for low task utilization

[jp walters]

Hi Björn,

On Fri, Nov 21, 2014 at 7:40 AM, Björn Brandenburg <bbb at mpi-sws.org> wrote:

> Hi JP,
>
> thanks for reporting the issue. I'm not sure what's going on. Can you
> recreate the issue without the scripts (which are a black box to me) just
> by running & tracing a couple of rtspin instances?
>
>
Which traces are you interested in?  I can recreate the 3-task system
easily enough, and I can capture /dev/litmus/log.  I can also use feather
trace to get /dev/litmus/sched_traceX data.  Poking around the list, I've
seen recommendations to use the following to parse out the sched_trace logs:

https://github.com/brandenburg/sched-trace-tools

I pulled that down, and using the job_stats tool on the output from
st_trace, I'm left with plaintext data that looks like below (one for each
CPU).  I'm probably doing something wrong though.

root at ubuntu-qemu:~/non-script.data# cat simple0
# Task,   Job,     Period,   Response, DL Miss?,   Lateness,  Tardiness
# task NAME=<unknown> PID=3802 COST=0 PERIOD=0 CPU=-1
  3802,     9,          0,   30657540,        0,  -29342460,          0
  3802,    13,          0,   32616504,        0,  -27383496,          0
  3802,    39,          0,   30793032,        0,  -29206968,          0
  3802,    89,          0,   29259038,        0,  -30740962,          0
  3802,    93,          0,   29860287,        0,  -30139713,          0
  3802,    97,          0,   29485435,        0,  -30514565,          0
  3802,   101,          0,   30333923,        0,  -29666077,          0
  3802,   131,          0,   30122722,        0,  -29877278,          0
  3802,   135,          0,   30649661,        0,  -29350339,          0
# task NAME=<unknown> PID=3801 COST=0 PERIOD=0 CPU=-1

To be clear, what I'm doing is the following:

./rtspin 10 20 10 -w &
./rtspin 30 60 10 -w &
./rtspin 60 120 10 -w &
./release_ts

Then, in a second window, I simply run:
st_trace simple




> > On 20 Nov 2014, at 22:01, jp walters <mailinglistjp39 at gmail.com> wrote:
> >
> > root at ubuntu-qemu:~/experiment-scripts# ./parse_exps.py run-data/test1
> > Loading experiments...
> > Parsing data...
> >  0.00%
> > Writing csvs into parse-data...
> > Too little data to make csv files, printing results.
>
> What does the above line mean?
>
>
I was hoping someone on the list might know.  I assumed that this was
normal, given the small system that I used.

thanks,
JP


> > <ExpPoint-run-data/test1>
> >            block-avg:  Avg:     0.000  Max:     0.000  Min:     0.000
> Var:     0.000
> >            block-max:  Avg:     0.000  Max:     0.000  Min:     0.000
> Var:     0.000
> >           miss-ratio:  Avg:     1.000  Max:     1.000  Min:     1.000
> Var:     0.000
> >          record-loss:  Avg:     0.000  Max:     0.000  Min:     0.000
> Var:     0.000
> >             tard-avg:  Avg:     0.000  Max:     0.000  Min:     0.000
> Var:     0.000
> >             tard-max:  Avg:     0.000  Max:     0.000  Min:     0.000
> Var:     0.000
> >
> > The 1.0 miss-ratio seems odd to me, given the resources available and
> the workload described in the 3 tasks example.  Am I missing something?
>
> This indeed seems odd. Miss _all_ deadlines? Even under moderate overload
> you would expect *some* jobs to complete before their deadlines. At the
> very least, the first jobs to be scheduled should not miss their deadlines.
>
> This means either that LITMUS^RT is somehow horribly broken or that the
> experimental setup isn't quite doing what it's supposed to. Let's try to
> narrow this down by using a simpler setup.
>
> Thanks,
> Björn
>
>
> _______________________________________________
> litmus-dev mailing list
> litmus-dev at lists.litmus-rt.org
> https://lists.litmus-rt.org/listinfo/litmus-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.litmus-rt.org/pipermail/litmus-dev/attachments/20141121/66d791e3/attachment.html>