Date   
Re: Interrupt Moderation on Ethernet cards

Thorsten Miglus
 

Thorsten


As I have written; The effect of buffering and delayed bulk transfer is not a PCIe bridge problem only. It can occurr with network cards or components. This has been confirmed by Klaus-Peter Renner. I also wonder what the EDIMAX stick does as they advertise energy saving modes at low bitrates. Newer Linux distros even have a daemon for interrupt moderation. So this opens a new battle field with try and error. I would expect that delivering every packet as fast as possible to the TC client is the way to go. With HVS-2 as sent right now we have a very good test signal :-).


Cheers,
Ernst
===============================

Ernst,

actually I take HVS-2 with an EDIMAX stick. I have disabled power saving modes.
For BAS / HVS-1 I use the build in Ethernet port.
When switching off interrupt moderation I can clearly see that Tellicast takes more processor time.

http://www.ing-miglus.de/eumetcast/cpu-8-9.html http://www.ing-miglus.de/eumetcast/cpu-8-9.html


CPU 8 (green) run Tellicast BAS and HVS-1.
CPU 9 (blue) run Tellicast HVS-2

At 17:00 UTC I disabled interrupt Moderation. CPU 8 load increased accordingly.

Cheers,
Thorsten

Re: HVS/TBS

Ernst Lobsiger
 

Mike


I setup a GNU/Linux EUMETCast receiver with a TBS-6903 on a private network with ssh access, encrypted RSA/DSA key authentication NAT and firewall settings for TBSDTVs IP segment of CHINA network. So be assured I still work with them. With CrazyCat involved there has been recent progress. But no magic either ..


Cheers,
Ernst



---In MSG-1@..., <mikeg4cfz@...> wrote :

There you go Ernst now you see my problem, its a new TBS-6903 and new Dual Tuner Software, sorry no magic. Mike

Re: HVS/TBS

mikeg4cfz
 

There you go Ernst now you see my problem, its a new TBS-6903 and new Dual Tuner Software, sorry no magic. Mike

Re: Interrupt Moderation on Ethernet cards

Ernst Lobsiger
 

Thorsten


As I have written; The effect of buffering and delayed bulk transfer is not a PCIe bridge problem only. It can occurr with network cards or components. This has been confirmed by Klaus-Peter Renner. I also wonder what the EDIMAX stick does as they advertise energy saving modes at low bitrates. Newer Linux distros even have a daemon for interrupt moderation. So this opens a new battle field with try and error. I would expect that delivering every packet as fast as possible to the TC client is the way to go. With HVS-2 as sent right now we have a very good test signal :-).


Cheers,
Ernst



---In MSG-1@..., <thorsten.miglus@...> wrote :

Hi all,

by default Interrupt Moderation is enabled within the drivers of Ethernet cards.
This means some packets are collected before they are delivered by an interrupt to the CPU.
Maybe for EUMETCAST reception it is a better way to disable Interrupt Moderation?
Then every single packet will cause an interrupt and delivered immediately.

https://community.rti.com/kb/how-reduce-latency-windows-disabling-interrupt-coalescing https://community.rti.com/kb/how-reduce-latency-windows-disabling-interrupt-coalescing https://community.rti.com/kb/how-reduce-latency-windows-disabling-interrupt-coalescing https://community.rti.com/kb/how-reduce-latency-windows-disabling-interrupt-coalescing


Cheers,
Thorsten

Re: HVS/TBS

Ernst Lobsiger
 

Douglas,


My recommandation for the moment:


If for some reason you take HVS-1 only do *not* switch BAS MODCODS off
If you want best BAS reception (heavy rain) do switch HVS-1 MODCODS off


EUMETSAT is willing to help. But they also say that constant (or even minimum) bitrates is something of the past. HVS using VCM is still top of the list but some PCIe card driver sources have ancestors from early PCI.
And these two "BUSes" are actually *very* different. PCIe is more some kind of network than a classic BUS.


Cheers,
Ernst

---In MSG-1@..., <dsdeans@...> wrote :


================================================================================================

Ernst I don't disagree with any of your analysis, just merely pointing
out that the HVS-1 arrangement is such that I am not losing any data
although I only take OLCI EFR.
I did point out sometime ago, before all the HVS-2 issues came to light,
that HVS-1 on its own had some problems.

Appreciate the work you are doing with this to try and get the problems
resolved.
I would hope to continue using the 6903 as my SR1 for reasons unknown
does not perform well with HVS data.

Regards,
Douglas.
>

Interrupt Moderation on Ethernet cards

Thorsten Miglus
 

Hi all,

by default Interrupt Moderation is enabled within the drivers of Ethernet cards.
This means some packets are collected before they are delivered by an interrupt to the CPU.
Maybe for EUMETCAST reception it is a better way to disable Interrupt Moderation?
Then every single packet will cause an interrupt and delivered immediately.

https://community.rti.com/kb/how-reduce-latency-windows-disabling-interrupt-coalescing https://community.rti.com/kb/how-reduce-latency-windows-disabling-interrupt-coalescing


Cheers,
Thorsten

Re: HVS/TBS

Douglas Deans
 

------ Original Message ------
From: "ernst.lobsiger@... [MSG-1]" <MSG-1@...>
To: MSG-1@...
Sent: 31/03/2017 20:27:19
Subject: Re: Re[2]: [MSG-1] Re: HVS/TBS

Douglas


Did you check the HVS-1 log with your icon changing colours? The
announcement channel disconnected. As for the files lost there is a
difference as well. HVS-1 has 101 minutes revolution periods with broad
gaps where you only have some HVS-KA dummy files sometimes you probably
don't even take. So the problem that you might miss a file is only
every 101 minutes when OLCI dissemination starts. Check the
TCLogSummary in my report: Only two files missed but 352 reconnects
(because annoncement channel lost). HVS-2 had it's gaps every minute.
Thats why you lost far more files there. HVS-2 is working like that or
even worse again just now.


Cheers,
Ernst
==================================================================================================

Ernst I don't disagree with any of your analysis, just merely pointing
out that the HVS-1 arrangement is such that I am not losing any data
although I only take OLCI EFR.
I did point out sometime ago, before all the HVS-2 issues came to light,
that HVS-1 on its own had some problems.

Appreciate the work you are doing with this to try and get the problems
resolved.
I would hope to continue using the 6903 as my SR1 for reasons unknown
does not perform well with HVS data.

Regards,
Douglas.

Re: HVS/TBS

Ernst Lobsiger
 

Thorsten


Out posts crossed. That's what I explained two posts back. When OLCI files start every 101 minutes and some untaken HVS-KA was there at that moment you will not loose anything. But look at the WEB Interface (Overview) of HVS-2 just now!


Ernst



===============================

Mike, Ernst and Douglas,


please take into account the different behaviour of HVS-1 and HVS-2 data streams.
HVS-2 has many short gaps in the data stream.
HVS-1 has a long gap of about one hour with no data channel active and about one hour data delivery with no gap.
No missed packets will be reported by TelliCast on HVS-1 during the long gaps. Even if the announcement channel disconnects and reconnects. This can be seen in the TelliCast log file only.
Checking the missed and recovered packets counter may not be enough to say: everything is ok.
To have a look into the TelliCast log files is required too.

Cheers,
Thorsten

Re: HVS/TBS

Thorsten Miglus
 

I have been running the HVS-1 on its own with MODCOD set to 16 APSK2/3
on one tuner of the TBS 6903 and BS on the other tuner using 8PSK3/5.
Under those conditions I have to say that whilst I am seeing BDADataEx
occasionally turn green (on the HVS-1) I have no lost data and only a
few lost packets but all recovered.
It does seem to perform better than the HVS-2 scenario which has lost
data and tmp folders left.

I cannot try it with the TBS IP tool as the dual tuner software is still
not available (out to some for testing). So Mike is right about no lost
data.

I have been running like this for sometime as it seems pointless running
T2 without meaningful data and the above arrangement prevents BS losses
if the weather is bad and affects HVS.

Regards,
Douglas.

===============================

Mike, Ernst and Douglas,


please take into account the different behaviour of HVS-1 and HVS-2 data streams.
HVS-2 has many short gaps in the data stream.
HVS-1 has a long gap of about one hour with no data channel active and about one hour data delivery with no gap.
No missed packets will be reported by TelliCast on HVS-1 during the long gaps. Even if the announcement channel disconnects and reconnects. This can be seen in the TelliCast log file only.
Checking the missed and recovered packets counter may not be enough to say: everything is ok.
To have a look into the TelliCast log files is required too.

Cheers,
Thorsten

Re: HVS/TBS

Ernst Lobsiger
 

Sorry


Actually 3 HVS-1 files missed ..:-) I need stronger glasses.


Ernst



---In MSG-1@..., <ernst.lobsiger@...> wrote :

Douglas

.. Only two files missed but 352 reconnects (because annoncement channel lost). ...


---In MSG-1@... mailto:MSG-1@..., <dsdeans@...> wrote :

Re: HVS/TBS

Ernst Lobsiger
 

Douglas


Did you check the HVS-1 log with your icon changing colours? The announcement channel disconnected. As for the files lost there is a difference as well. HVS-1 has 101 minutes revolution periods with broad gaps where you only have some HVS-KA dummy files sometimes you probably don't even take. So the problem that you might miss a file is only every 101 minutes when OLCI dissemination starts. Check the TCLogSummary in my report: Only two files missed but 352 reconnects (because annoncement channel lost). HVS-2 had it's gaps every minute. Thats why you lost far more files there. HVS-2 is working like that or even worse again just now.


Cheers,
Ernst



---In MSG-1@..., <dsdeans@...> wrote :

------ Original Message ------
From: "ernst.lobsiger@... mailto:ernst.lobsiger@... [MSG-1]" <MSG-1@... mailto:MSG-1@...>
To: MSG-1@... mailto:MSG-1@...
Sent: 31/03/2017 19:06:41
Subject: Re: [MSG-1] Re: HVS/TBS

>Mike
>
>
> HVS-1 on ist own means you have to set MODCOD 16APSK2/3 on the tuner
>you run the HVS-1 client. If you only stop the BAS client the 8PSK3/5
>traffic is still there and keeps the DMA buffers going. Arne made the
>test with his TBS-6983 as well. So unless you have an all new TBS card
>or driver it's again some of your magic ...
>
>
> Cheers,
> Ernst
==============================================================================================

I have been running the HVS-1 on its own with MODCOD set to 16 APSK2/3
on one tuner of the TBS 6903 and BS on the other tuner using 8PSK3/5.
Under those conditions I have to say that whilst I am seeing BDADataEx
occasionally turn green (on the HVS-1) I have no lost data and only a
few lost packets but all recovered.
It does seem to perform better than the HVS-2 scenario which has lost
data and tmp folders left.

I cannot try it with the TBS IP tool as the dual tuner software is still
not available (out to some for testing). So Mike is right about no lost
data.

I have been running like this for sometime as it seems pointless running
T2 without meaningful data and the above arrangement prevents BS losses
if the weather is bad and affects HVS.

Regards,
Douglas.
>

Re: HVS/TBS

Douglas Deans
 

------ Original Message ------
From: "ernst.lobsiger@... [MSG-1]" <MSG-1@...>
To: MSG-1@...
Sent: 31/03/2017 19:06:41
Subject: Re: [MSG-1] Re: HVS/TBS

Mike


HVS-1 on ist own means you have to set MODCOD 16APSK2/3 on the tuner
you run the HVS-1 client. If you only stop the BAS client the 8PSK3/5
traffic is still there and keeps the DMA buffers going. Arne made the
test with his TBS-6983 as well. So unless you have an all new TBS card
or driver it's again some of your magic ...


Cheers,
Ernst
==============================================================================================

I have been running the HVS-1 on its own with MODCOD set to 16 APSK2/3
on one tuner of the TBS 6903 and BS on the other tuner using 8PSK3/5.
Under those conditions I have to say that whilst I am seeing BDADataEx
occasionally turn green (on the HVS-1) I have no lost data and only a
few lost packets but all recovered.
It does seem to perform better than the HVS-2 scenario which has lost
data and tmp folders left.

I cannot try it with the TBS IP tool as the dual tuner software is still
not available (out to some for testing). So Mike is right about no lost
data.

I have been running like this for sometime as it seems pointless running
T2 without meaningful data and the above arrangement prevents BS losses
if the weather is bad and affects HVS.

Regards,
Douglas.

Re: HVS/TBS

Ernst Lobsiger
 

Mike


HVS-1 on ist own means you have to set MODCOD 16APSK2/3 on the tuner you run the HVS-1 client. If you only stop the BAS client the 8PSK3/5 traffic is still there and keeps the DMA buffers going. Arne made the test with his TBS-6983 as well. So unless you have an all new TBS card or driver it's again some of your magic ...


Cheers,
Ernst



---In MSG-1@..., <mikeg4cfz@...> wrote :

Ernst, excellent report and description of the problems, I admit to not being so Technical as some but I understood most of the details, but here I have a problem, I am using TBS Dual Tuner Software, and I am able to run HVS-1 on its own on one Tuner, or run both BS on Tuner A and HVS-1 on Tuner B, in fact I have been running HVS-1 on its own all day on Tuner A without a single packet loss and no BS, I have no explanation that's just the way it is. Regards, Mike.

Re: HVS/TBS

mikeg4cfz
 

Ernst, excellent report and description of the problems, I admit to not being so Technical as some but I understood most of the details, but here I have a problem, I am using TBS Dual Tuner Software, and I am able to run HVS-1 on its own on one Tuner, or run both BS on Tuner A and HVS-1 on Tuner B, in fact I have been running HVS-1 on its own all day on Tuner A without a single packet loss and no BS, I have no explanation that's just the way it is. Regards, Mike.

Re: HVS/TBS

Ernst Lobsiger
 

Dear All,


As promised I updated my paper and made an official OPS request to improve reception for TBS products.
I also corrected some minor mistakes, added latest results and some more pointers. I skipped 3 pages of announcement channel lost messages. You can download the final and slightly renamed version here:



http://5.153.126.239:85/pdf/Reconfigure_EUMETCast.pdf http://5.153.126.239:85/pdf/Reconfigure_EUMETCast.pdf


Here is my request:




Request to improve EUMETCast to better support TBS DVB devices


Dear OPS


For month now the ever broadening group of EUMETCast stations using
COTS receivers from TBS (now also emerging from Digital Devices) has
struggled with receiving the second transponder T2 still in test mode.


In a common effort we have pinned down the problem. Our analysis shows
that reconfiguration at the EUMETCast UpLink side can solve the problem.
It also shows that this is not a problem of T2 alone but has existed
on T1 from day one though hidden by the BAS dissemination schedule.


On behalf of all users receiving EUMETCast with devices different from
DVB routers (EUMETSAT has a couple of such devices marked as supported on
the homepage) I ask you to study the attached report and take appropriate
measures. This should enable all stations to receive T2 by July 1st 2017.
It could even improve the reception quality for BAS and HVS-1 at large.


Thank you for your help.


Best Regards,
Ernst Lobsiger





---In MSG-1@..., <ernst.lobsiger@...> wrote :

Alan,


. . I willl update my paper ... and then make an official request to EUMETSAT via OPS.


Best Regards,
Ernst

Re: Data drop yesterday....

David J Taylor
 

Hi David,

MODIS-L1 data dropped by 3 GB. More then 200 Files less than the day before.

Cheers,
Thorsten
===========================

Thanks, Thorsten. Let's hope it's back today. Those satellites are quite
old now....

Cheers,
David
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv

Re: Connecting

David J Taylor
 

Bonjours à tous,

Thank you very much David and Ernst for your help. Everything was normal and
yet HVS remained "connecting".

I finally reloaded and installed the SR1 configuration again and without
changing anything else everything is back in order and data is coming in
again !

Thanks again

Best regards
Gerard
===============================

Gerard,

That's good news. Sounds like you were missing a PID, or something.

73,
David GM8ARV
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv

Re: Question about TBS Pids

David J Taylor
 

Douglas, are you saying that Transponder 2 is on a different frequency. I was under the impression that everything is on 11263 mhz/ 3300. I am receiving both basic and HVS-1 on that frequency. I cannot find any info to the contrary.
===============================

Greg,

A quick summary of PIDs and file settings:

PIDs
610 0x0262 HVS-2 test service
611 0x0263 HVS-2 test service

======== cast-client_hvs-2.ini ========
[announcement_channel]
name=TSL-HVS-2
address=224.223.224.223:4711

======== cast-client-channels_hvs-2.ini ========
[channel]
name=*
target_directory=&#92;EUMETCast&#92;received&#92;hvs-2&#92;default
tmp_directory=&#92;EUMETCast&#92;tmp&#92;hvs-2

Cheers,
David
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv

Re: Question about TBS Pids

David J Taylor
 

Douglas, are you saying that Transponder 2 is on a different frequency. I was under the impression that everything is on 11263 mhz/ 3300. I am receiving both basic and HVS-1 on that frequency. I cannot find any info to the contrary.
===============================

Greg,

Perhaps you missed my posting:

http://www.satsignal.eu/wxsat/dvb-s2/T1-T2.html#setup

Cheers,
David
--
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv

Re: Question about TBS Pids

Douglas Deans
 

------ Original Message ------
From: "greggrant890@... [MSG-1]" <MSG-1@...>
To: MSG-1@...
Sent: 31/03/2017 13:12:40
Subject: Re: [MSG-1] Question about TBS Pids

Douglas, are you saying that Transponder 2 is on a different frequency.
I was under the impression that everything is on 11263 mhz/ 3300. I am
receiving both basic and HVS-1 on that frequency. I cannot find any
info to the contrary.
===================================================

Yes Transponder 2 is on 11378.500 MHz.
Please see :-

http://www.eumetsat.int/website/home/TechnicalBulletins/EUMETCast/DAT_3340783.html

For the moment you will see it has a different symbol rate and MODCOD
but that will likely change when Eumetsat get the whole transponder in
July.
Only test data is being sent and although it does say GOES 16 data in Q1
2017 that clearly is not now happening.

Hope that helps.
Regards,
Douglas.