Date   
Re: Tellicast 2.14.5-1 100% CPU

George Sz
 

Ernst,

Thanks for your tips. In the end, I managed to resolve the issue, here is a breakdown of what I tried and what worked or didn't.

None of the following worked:
- deployed Devuan ASCII
- turned off all CPU power saving features like EIST and C states
- disabled IPv6 (I'm getting it natively from my ISP)
- disconnected computer from network (complete offline mode)
- fixed SACSrv init script
- disabled SACSrv since it doesn't seem to be needed at all (?)
- ran 'pcscd -f -d' (foreground, debug to stdout... no errors there)
- ran tellicast manually with -v (did not output anything useful)
- did not open webif (thought it might be some sort of memory leak)

What did work:
+ installed multiarch i386
+ installed 32 bit pcscd
+ installed 32 bit drivers and libs
+ installed 32 bit tellicast

So far, this setup has been running for the past 6 hours or so, on my i5-4570, with no CPU hogging. There must be something wrong with the 64 bit version, since I'm always getting this issue, even with the default .cfg that comes with the installer, practically on any Linux distro, no matter what hardware it's running on. Last night I thought I got the 64 bit version working. It was running for about an hour and the CPU issue got triggered.

Problem is, this seems to be completely random and I can't trigger the issue manually during normal operation. It just happens. I also observed that running the client with no EKU, then plugging it in after manually starting tellicast immediately triggers the CPU issue. After that, tellicast stays at 100% CPU, even when unplugging EKU.

Regards,
George

Re: Tellicast 2.14.5-1 100% CPU

Ernst Lobsiger
 

George,

I have never used htop before. Now I installed it to compare with your htop output.
Your pcscd has --foreground --auto-exit switches. Is this part of your debugging?

On my receiver Io htop shows no switches for pcscd. The manual page of pcscd states under --auto-exit
"pcscd will quit after 60 seconds of inactivity. This is used when pcscd os started on demand by systemd".

I think pcscd must be started once for good and should not be started on demand by systemd. This could be a result of the missing fix in /etc/init.d/SACSrv?
Maybe you make the fix in /etc/SACSrv (see my last post). And make sure SACSrv is automatically started when booting. Without systemd I'd use insserv ...

Regards,
Ernst

Re: Tellicast 2.14.5-1 100% CPU

Ernst Lobsiger
 

George,

I forgot one thing to mention. In script /etc/init.d/SACSrv I had to add an INFO line to make it standard:
....
# Required-Start:    pcscd
# Required-Stop:             <--- This line was missing but must be there even if no run levels mentionned!!
# Default-Start:       2 3 4 5
......

Maybe this starts more than one SACSrv or does not work with systemd out of the box?
I should have dropped a line to OPS or K.-P. but didn't either. :-(. Maybe it's worth a try?

Cheers,
Ernst

Re: Tellicast 2.14.5-1 100% CPU

Ernst Lobsiger
 

George,

the main difference is probably that Devuan ASCII is systemd free. I will never get used to this redhat stuff. But I'am not sure Devuan developers can keep their effort up for years to come. I also use my own dvb-eumetcast script with 3 dummy interfaces (... unlike EUMETSAT's also working when stopped). Appart from this I do not see why TelliCast  should not run with Ubuntu. If you use your 10 years old MB disable all energy saving C-states in the BIOS if that can be done.

Good luck,
Ernst

Re: Tellicast 2.14.5-1 100% CPU

George Sz
 

I'll try the distro that you're using and see what happens. For reference, on my Ubuntu Server install it looks like this:

Re: Tellicast 2.14.5-1 100% CPU

Ernst Lobsiger
 

George,

all I know is described in my attachment of message #28250. I installed the SafenetAuthenticationClient-core-9.0.43-0_amd64.deb.
It's now a while I haven't run this TelliCast receiver Io. I just restarted it now and you should see it on http://185.74.120.175:81
It will have a couple of lost packets though due to the latest TBS-6903X card with still experimental driver it uses.

Cheers,
Ernst

Re: Tellicast 2.14.5-1 100% CPU

George Sz
 

I did. tc-cast-client -k checks out fine, but as the client is running and it interacts with the EKU, sooner or later it will hog the CPU, even if it's not receiving any data (dvb-eumetcast stopped). The other PC I mentioned is of a much older architecture. It's an E5300 running in some 10yr old Gigabyte motherboard. My main eumetcast PC is a Lenovo i5 Haswell. Makes no difference, same CPU hog. It seems Tellicast is bombarding pcscd with CMD_GET_READERS_STATE and somehow that makes it go haywire. I don't remember the old client doing this but unfortunately I don't have a backup of that.

I also learned there is a newer version of the Safenet software but I did not find it anywhere. It's available from the manufacturer but it's behind a wall. Maybe that would help, I don't know.

Re: Tellicast 2.14.5-1 100% CPU

Ernst Lobsiger
 

George,

did you also install the new 64Bit EKU software? See my message #28250.

Ernst

Re: Tellicast 2.14.5-1 100% CPU

George Sz
 

I narrowed this down to a problem with my hw key. When running pcscd in debug mode, the following repeats over and over when tellicast is running.

00000029 [139713218291456] winscard_svc.c:361:ContextThread() Received command: CMD_STOP_WAITING_READER_STATE_CHANGE from client 8
00000008 [139713218291456] winscard_svc.c:440:ContextThread() CMD_STOP_WAITING_READER_STATE_CHANGE rv=0x0 for client 8
00000028 [139713218291456] winscard_svc.c:361:ContextThread() Received command: CMD_WAIT_READER_STATE_CHANGE from client 8
00000006 [139713218291456] winscard_svc.c:832:MSGSendReaderStates() Send reader states: 8

I don't remember this happening with the old client/driver combo. I tried a different distro on a different computer and the result is exactly the same. Without the key plugged in, tellicast is running normally, but it's not much of a solution. I'm clueless at the moment, really.

I can't believe Eumetsat hasn't noticed this, or maybe they're just running some ancient version of Linux? I'm running Ubuntu Server 18.04.2 LTS and Fedora 30.

Re: Metop C and GOES-17

stevens312
 

Thank You Douglas,picked up some virus which knocked us both down, ok will start looking into that now. Regards, Mike.

Re: Metop C and GOES-17

Douglas Deans
 

On 11/05/2019 11:23, stevens312 wrote:
Hi Guys, sorry if this is bit behind the times but trying to receive the above data and not sure where it is located, due to some health problems been a bit out of it lately so require some up-date on whats new, had the EKU updated by EUMETSAT, any advice gratefully received. Regards to all, Mike G4CFZ.
=========================================================================

Hello Mike,
Both those services are now on HVS-1.

Metop-C AVHRR is channel E1H-EPS-10 and
GOES 17 is channel E1H-TPG-3. Unlike GOES 16 which has 15 minute data, the GOES 17 is ½ hourly intervals.

There are other Metop services on HVS-1 such as GRAS etc., but I assume those will be of little interest.

Hope that helps and hope you are now keeping better.

Best regards,
Douglas.

Metop C and GOES-17

stevens312
 

Hi Guys, sorry if this is bit behind the times but trying to receive the above data and not sure where it is located, due to some health problems been a bit out of it lately so require some up-date on whats new, had the EKU updated by EUMETSAT, any advice gratefully received. Regards to all, Mike G4CFZ.

Re: New version EUMETCastView 1.4.6

Hugo
 

Hello George,

As already mentioned by Ernst, the I-band of VIIRS is not available in Eumetcast. Maybe in the future ...

Kind regards,

Hugo

Re: New version EUMETCastView 1.4.6

Ernst Lobsiger
 

George,

There is AFAIK no VIIRS I-Bands on EUMETCast so far. That's probably why Hugo does not (yet) support it.
If you are looking for a rather complete l1b VIIRS support including many composites you could try PyTROLL.

It's still in development but you can already do wonderful things (like some we just recently found out ...).

Regards,
Ernst

Tellicast 2.14.5-1 100% CPU

George Sz
 

Hello,

I've upgraded to the latest, now finally 64 bit software package from Eumetsat and I immediately ran into an issue. After about 10-15 minutes of normal operation, the process shoots up to 100% CPU usage and it stays that way. It even triggers the built-in watchdog, saying it exceeded the 95% CPU threshold. I disabled this watchdog and the software works for now but this is not optimal (especially not for power usage).

Has anybody ran into this issue? I was looking at htop but it doesn't show any useful info. If anybody experienced this error and found a fix, please let me know.

Thanks,
George

Re: New version EUMETCastView 1.4.6

George Sz
 

Hello,

Thanks for the release.

I've been using your software now and then, mostly for VIIRS viewing. Would you be able to add I-band processing capability to your software? I'd be willing to donate for that cause.

Regards,
George

Re: 1.2m DISH

Ernst Lobsiger
 

Robert,

the Gibertini OP-125L is a good dish though I had some mechanical problems with elevation settings.
To tranport such a dish on the roof of a custom car is quite an adventure to say the least. Speed limit
whithout wind is 45 MPH whatever your fixation. I have done that some 150 Miles on highways and
know what I'am talking about ...

Regards,
Ernst

Re: 1.2m DISH

Brian
 

Have a Gibertini not used in years and then only briefly as problem was LNB, it currently lives in the garage!
Was keeping it for HVS-1 & 2 but 1M dish works perfectly here except under heavy rain.

Location is near Harwich.

73's
Brian

1.2m DISH

Robert
 

Hi
I am looking for a 1.2m dish for receiving HVS-2.  I do want it both ways!! ie not too expensive but a good quality dish.

Does anyone have one for sale or can you suggest where is a good place to purchase one.   

I can collect if it is not too far away.  My location is Rudgwick nr Horsham

Many thanks ans 73s

Robert
G4ZEJ

Re: file problem

David J Taylor
 

From: Heinz Meschnark

Hello!
I am using eumetsat.
As you can see I can recieve the files partially.
When I enter eumetsat.bas, the icon is dark.
Then I get the message "warning reset" or "error".
I am using the Msg Manager V3.1.10.1268 and cannot see this software working.
Please can you help
=================================

Heinz,

Thanks for posting the screen-shots.

Looking at the TelliCast Statistics page, I see a non-zero value for Missed Packets, and for Recovered Packets. For receiving just the Basic Service (which I would advise to start with) I would expect a zero value for both numbers as you are located in Austria. The exception might be if you now have the High Volume Service (HVS-1) enabled as that does need more signal (perhaps 80 cm in Austria?).

But as Gerard notes, it may just be getting that received data path correct.

73,
David GM8ARV
--
SatSignal Software - Quality software for you
Web: http://www.satsignal.eu
Email: david-taylor@...
Twitter: @gm8arv