[Etherlab-users] Data loss in DLS

[Andreas Stewering-Bone]

Dear Amos,

There can be severall different possibilities:


First of all what is behind the high time difference. The DLS stores
data in blocks with a start timestamp a count of blocks and a step
time.If start + count * step > next start you will get this failure.

This can be caused by:
-RT properties, please check your system with cyclictest (rt-test
suite)!
- Check your process cyclic time and execution time!
- What the normal system load of the system (top)?
- What is the normal IO load of the system (iotop)?
- Wat is your DLS blocksize. A small blocksize compared to your data
frequency will cause high HDD load. My favorite settings freq/count
blocks =1, means once a second a block is written per channel.
- If you have a simulink based model, NTP sync will not be a problem,
because there is an internal time controller in the model. 


Best regards

Andreas


Am Freitag, dem 17.03.2023 um 15:56 +0000 schrieb Merkel, Amos:
> Hi,
>
> Thanks for the hint! Both DLS and the control job run locally on the
> same machine, so I expect they are using the same time source. I
> should have mentioned that in my first post, sorry.  
>
> However, my ntp was indeed not synchronised, as I recently turned it
> off as part of my troubleshooting. I did not notice any impact from
> turning it off on system level.
> I turned it on again now, but do not expect a change in behaviour, as
> it should have been running when the errors first occurred.
>
> Is there an active synchronisation of the time beyond the system
> clock?
>
> Thank you & all the best
> Amos
>
> --
> Amos Merkel, M.Sc.
> Gruppenleiter
> Digitalisierung
>
> ifas – Institut für fluidtechnische Antriebe und Systeme
> RWTH Aachen University
> Campus-Boulevard 30
> 52074 Aachen
>
> Tel: +49 241 80-47737
> Fax: +49 241 80-647712
>
> www.ifas.rwth-aachen.de | Twitter | Facebook | YouTube | LinkedIn
>
>
>
>
>
>
>
> -----Ursprüngliche Nachricht-----
> Von: Sebastien BLANCHET <blanchet at iram.fr>
> Gesendet: Freitag, 17. März 2023 15:38
> An: Merkel, Amos <Amos.Merkel at ifas.rwth-aachen.de>;
> etherlab-users at etherlab.org
> Betreff: Re: [Etherlab-users] Data loss in DLS
>
> Hi,
>
> Could you check that
> - DLS and the data sources are synchronized with NTP
> - your NTP server works
>
> --
> Sebastien BLANCHET
>
>
> On 17/03/2023 14:56, Merkel, Amos wrote:
> > Hi everyone,
> >
> > we are experiencing weird data losses on the DLS. Everything works
> > as
> > expected for a couple of hours. Then, suddenly, there is a message
> > in
> > the logs from dlsd_logger declaring an unexpected high time
> > difference
> > in a number of channels followed by a message that the msr protocol
> > was
> > reset by the peer and dls stops logging with error code -2. Logging
> > is
> > continued after the preset 30 seconds, but typically crashes again
> > shortly after.
> >
> > I am quite at loss how to proceed, there is nothing apparently
> > wrong –
> > the same system set-up has worked up to now, there is nothing
> > unusual in
> > the logs and system load (disc, cpu, network, ram) appears normal.
> > The
> > error appears on several systems, the only common thing appears to
> > be a
> > certain hardware configuration (certain somewhat older fujitsu
> > desktops). Unfortunately, I do not have enough other hardware at
> > hand to
> > really support that theory.
> >
> > We work with openSUSE and installed our etherlab environment from
> > the
> > science:EtherLab repository using the generic network driver. We
> > see the
> > error on 15.2 and 15.3 systems, but haven’t managed to check 15.4
> > yet.
> >
> > Has anybody experienced similar behaviour before and has a hint on
> > possible causes or fixes?
> >
> > All the best,
> >
> > Amos
> >
> > --
> > Amos Merkel, M.Sc.
> > Group leader
> > Digitalization
> >
> > ifas – Institute for Fluid Power Drives and Systems
> > RWTH Aachen University
> > Campus-Boulevard 30
> > 52074 Aachen
> >
> > Tel: +49 241 80-47737
> > Fax: +49 241 80-647712
> >
> > www.ifas.rwth-aachen.de <http://www.ifas.rwth-aachen.de/>| Twitter
> > <https://twitter.com/IFAS_RWTH>| Facebook
> > <https://www.facebook.com/IFAS-Institut-f%C3%BCr-fluidtechnische-Antriebe-und-Steuerungen-260714043957873/
> > >| YouTube
> > <https://www.youtube.com/channel/UC0vIq8jtra8E_--Sx8DALkQ>|
> > LinkedIn <https://www.linkedin.com/company/ifas-rwth/>
> >
> > **
> >
> >         
> >
> >
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.etherlab.org/pipermail/etherlab-users/attachments/20230320/a6116cc4/attachment.htm>