Re: [wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 RC2
So 'm back with some more information. It's not a HF related issue, it never occurred during TX(everything is proper grounded, shielded cables etc. ), only during RX. Sometimes I get an error message like "Data not feed to soundcard fast enough"(during TX). But this does not stop the decode process nor the waterfall. I've been testing with some kernel parameters, these problems are Raspberry Pi related. One can effect the acceleration of different USB types (iso or asynchronous, interrupt etc.), since then I did not see this problem so often. But there should be a better error handling between different modules of wstjx. I can not trace the problem because no debug data is available. During my test I found something interesting for other Raspberry Pi or SBC users. All data, during program execution, is written to /tmp. This is a big problem because /tmp is part of the root- filesystems and this is located on a sdcard. This stresses sdcard over normal use -> defect. So one has to use a tmpfilesystem to store everything in RAM (how big???) or on adiffertent storage location. Because I did not find any config file to specify the /tmp location. 73 Frank DB5FP > I had this if I used more than 5 watts on 40mtrs, it used to knock > the cat control out as well (ic-9100). Bought a decent screened USB > lead and no issues now. I assume the RF is still there but it is no > longer picked up. It is funny that we spend a fortune on radio and > antenna but use the USB lead from a pound shop. > > Richard G7OED > > From: Black Michael via wsjt-devel > Sent: Saturday, March 24, 2018 12:34 PM > To: wsjt-devel@lists.sourceforge.net > Cc: Black Michael > Subject: Re: [wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 > RC2 > > That sounds like your audio stream froze. > Possible causes: > > #1 RF in the shack. > #2 USB going to sleep -- this should only happen if PC is left > unattended. > > This is happening while your operating, right? In that case RF is > the likely culprit. Does happen when you transmit? Lots of potential > sources can cause this, USB cables, ethernet cables, computer cards. > We had one guy whose rotor caused it even though it wasn't plugged > into anything but power. > > > de Mike W9MDB > > > > > > On Saturday, March 24, 2018, 3:35:53 AM CDT, FRANK DB5FP > <db...@gmx.de> wrote: > > > > > > > Hi Bill, > > If the WSJT-X Rx level > > indicator is not showing a level *and* the waterfall stops > > scrolling, then it is fairly certain that the operating system is > > no longer delivering audio samples to the application. > > The indicator is showing a level but it's not moving and the > progessbar(15sec.) is still moving. > > 73s > > Frank > > DB5FP > > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > ___ > wsjt-devel mailing list > wsjt-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/wsjt-devel > > > > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > > > > ___ > wsjt-devel mailing list > wsjt-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/wsjt-devel > > > --- > This email has been checked for viruses by Avast antivirus software. > https://www.avast.com/antivirus -- Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 RC2
> Hi Bill, If the WSJT-X Rx level > indicator is not showing a level *and* the waterfall stops scrolling, > then it is fairly certain that the operating system is no longer > delivering audio samples to the application. The indicator is showing a level but it's not moving and the progessbar(15sec.) is still moving. 73s Frank DB5FP -- Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 RC2
> On 22/03/2018 18:42, FRANK DB5FP wrote: > > Hello, > > > > I'm back with some more Information : > > > > - the soundcard is not a cheapie:-) It's a Terratec Aureon > > I've used it several years for decoding NOAA sats, without > > any notice, even under linux and on RPI2 together with wxtoimg. > > - the problem looks like an RPI USB problem. > > > >I have been searching the internet and found "dwc_otg" > >problems.there are many forum entries and debug hints like > > kernel command line options etc., but at least nothing will help > > solving that usb-driver (that is my conclusion). > > - Main problem is of cours that linux-usb handling will loose some > >audiodata -> this causes the /usr/bin/JT9 process to stop working > >correctly (lower cpu usage than normal), no decode no waterfall > >This stops processing audiodata even after a short wile > >audiodatat is back again > > > > --> wsjt-x gui should supervise JT9 process to observe any > > irregularities , like it does when JT9 process is shut down > > manually. > > > > 73 > > > > Frank > > DB5FP > > Hi Frank, > > a couple of clarifications. > > The jt9 process does not directly handle audio samples, they are read > and written by a thread of the wsjtx process. > > The jt9 process is monitored by the wsjtx process and failures are > reported via a message box including the reason for failure. > > If the source of sound samples stops delivering samples; the wsjtx > waterfall and decoding will also stop. > > 73 > Bill > G4WJS. > Hi Bill, interesting Information, because : Soundcard has an indicator LED, LED ON means, power ok. Flashing means data is being read from the soundcard. And the LED is flashing(all the time)even when waterfall and decode stopped. -> wsjtx reads data from soundcard??? but does not pass data to jt9 ??? No Error Messages are generated in wsjtx. Behavior is the same like I'm pulling USB cord while wsjtx is running-> no message is generated and decode and waterfall is stopped. 73 Frank DB5FP -- Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 RC2
Hello, I'm back with some more Information : - the soundcard is not a cheapie :-) It's a Terratec Aureon I've used it several years for decoding NOAA sats, without any notice, even under linux and on RPI2 together with wxtoimg. - the problem looks like an RPI USB problem. I have been searching the internet and found "dwc_otg" problems.there are many forum entries and debug hints like kernel command line options etc., but at least nothing will help solving that usb-driver (that is my conclusion). - Main problem is of cours that linux-usb handling will loose some audiodata -> this causes the /usr/bin/JT9 process to stop working correctly (lower cpu usage than normal), no decode no waterfall This stops processing audiodata even after a short wile audiodatat is back again --> wsjt-x gui should supervise JT9 process to observe any irregularities , like it does when JT9 process is shut down manually. 73 Frank DB5FP > Hi Frank, > I've been running wsjt-x 1.8.0 on a rpi3 for weeks now, using a neat > qrp-labs receiver variant. FT8 decodes takes about 5-10 seconds on a > busy band. Speculations follow: > What kind of sound card are you running ? I'm guessing USB. If so then > there's certainly a possibility of USB problems or the soundcard > itself, certainly if it's a cheapie. This should however produce > messages in dmesg/messages. > My setup uses a audio-injector zero that connects through the 40-pin > header and uses the i2s protocol. It has been working nicely with > just a few hiccups reported in dmesg, but never stopped running or > required any action. I think it would be a good test to try changing > at least the sound card, or other stuff to analyze in which direction > the problem may be. One problem I have is that there's a lot of > orphaned .wav-files left after decode, it's a problem because I > use /tmp (50MB) as the storage for these to avoid writing lots of > stuff to the microSD. The workaround is a cleanup-script executed > periodically with crontab. > > 73's and good luck -- Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 RC2
Hi Greg, sorry, I did not mention this detail. Memory usage is always about 50% (of 1GB). 73 Frank, DB5PF Hi Frank, In this case, processor utilization is probably less of a concern than memory in terms of overall performance. The RIP2 is short to begin with for "normal" desktop activities. I believe the Pi2 B/(x) models came with ~1GB; even less on older versions. While you are watching HTOP, keep an eye on mem-usage, and when/how-much. You can set the sort order by memory or any number of filters using HTOP. 73's Greg, KI7MT -Original Message- From: FRANK DB5FP [mailto:db...@gmx.de] Sent: Tuesday, March 20, 2018 2:25 PM To: wsjt-devel@lists.sourceforge.net Subject: Re: [wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 RC2 Hi Joe, sorry first e-Mail did not pass my filter, so it went to spam...and I send a new one. At first, green Monitor light does NOT go out. Everything looks fine. But waterfall stops and no more messages are decoded. Sometimes it is possible to recover the program by pressing the Monitor buton twice. I tried 1.80 and 1.90 rc2only one instance. Both version have the same issue. I'm running FT8 on different bands and report the messages to pskreporter. I tried to get some more information by inspecting the processes with "htop" . It looks like the JT9 process is running but it's more or less idling 2.4% of CPU usage. During decode it's about 100% (1 core). During normal idle it's about 11%. If I shut down the JT9 process by kill-command and afterwards killing rest of the application, I can restart the application. By ending the program with ALT-F4 oder File-Exit I can not restart because of zombie processes on the message accrues that there would be a lock file. dmesg and /var/log/mesages don't show any problems. The problem is independent of any other program running beside wsjtx. But normally I use gpredict in the same time. Is there any error-loging possibility in wsjtx ? 73 Frank DB5FP > Hi Frank, > > On 3/17/2018 6:36 AM, FRANK DB5FP wrote: > > on my Raspberry Pi2 Monitoring Mode stopps suddenly. > > OS Version : Debian > > Kernel Linux version 4.14.18-v7+ > > > > Suddenly the Waterfall display stops and no messages are decoded BUT > > progressbar is still running. > > Rig controll ist still working. > > > > Any hints? > > You have reported this problem twice now. It might help to know what > WSJT-X (or possibly some other program running in your RasPi2 ??) is > doing when the green Monitor light goes out. What mode are you > running? What other processes are taking CPU time? > > Your system is, of course, in the "bare minimum" category for running > WSJT-X. Have you communicated with others who run WSJT-X on a > Raspberry Pi2 ? > > -- 73, Joe, K1JT -- Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
Re: [wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 RC2
Hi Joe, sorry first e-Mail did not pass my filter, so it went to spam...and I send a new one. At first, green Monitor light does NOT go out. Everything looks fine. But waterfall stops and no more messages are decoded. Sometimes it is possible to recover the program by pressing the Monitor buton twice. I tried 1.80 and 1.90 rc2only one instance. Both version have the same issue. I'm running FT8 on different bands and report the messages to pskreporter. I tried to get some more information by inspecting the processes with "htop" . It looks like the JT9 process is running but it's more or less idling 2.4% of CPU usage. During decode it's about 100% (1 core). During normal idle it's about 11%. If I shut down the JT9 process by kill-command and afterwards killing rest of the application, I can restart the application. By ending the program with ALT-F4 oder File-Exit I can not restart because of zombie processes on the message accrues that there would be a lock file. dmesg and /var/log/mesages don't show any problems. The problem is independent of any other program running beside wsjtx. But normally I use gpredict in the same time. Is there any error-loging possibility in wsjtx ? 73 Frank DB5FP > Hi Frank, > > On 3/17/2018 6:36 AM, FRANK DB5FP wrote: > > on my Raspberry Pi2 Monitoring Mode stopps suddenly. > > OS Version : Debian > > Kernel Linux version 4.14.18-v7+ > > > > Suddenly the Waterfall display stops and no messages are decoded BUT > > progressbar is still running. > > Rig controll ist still working. > > > > Any hints? > > You have reported this problem twice now. It might help to know what > WSJT-X (or possibly some other program running in your RasPi2 ??) is > doing when the green Monitor light goes out. What mode are you > running? What other processes are taking CPU time? > > Your system is, of course, in the "bare minimum" category for running > WSJT-X. Have you communicated with others who run WSJT-X on a > Raspberry Pi2 ? > > -- 73, Joe, K1JT > > -- > Check out the vibrant tech community on one of the world's most > engaging tech sites, Slashdot.org! http://sdm.link/slashdot > ___ > wsjt-devel mailing list > wsjt-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/wsjt-devel -- Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
[wsjt-devel] Monitoring stops suddenly in 1.80 and 1.90 RC2
Hallo, on my Raspberry Pi2 Monitoring Mode stopps suddenly. OS Version : Debian Kernel Linux version 4.14.18-v7+ Suddenly the Waterfall display stops and no messages are decoded BUT progressbar is still running. Rig controll ist still working. Any hints? 73 Frank DB5FP -- Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot ___ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
