search

RidgeRun / gst-shark

GstShark is a front-end for GStreamer traces
GNU Lesser General Public License v2.1
95 stars 45 forks source link

gstshark not working, datastream and metadata bad? #69

Closed tmbuckley closed 4 years ago

tmbuckley commented 4 years ago

./gstshark-plot ~/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50 [warning] Ill-shapen or missing "/* CTF x.y" header for text-only metadata. [warning] Unsupported CTF specification version 127.0. Trying anyway. [error] at line 1: token "": syntax error, unexpected $end

[error] Error creating AST [warning] Unable to open trace metadata for path "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50". [warning] [Context] Cannot open_trace of format ctf at path /home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50. [warning] [Context] cannot open trace "/home/tbuckley/tmp/seattle_tx2/VideoSystehark_2020-05-06_14:54:50 for reading. [error] Cannot open any trace for reading.

[error] opening trace "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-

[error] none of the specified trace paths could be opened.

michaelgruner commented 4 years ago

Can you share your trace files?

On May 6, 2020, at 14:13, tmbuckley notifications@github.com wrote:

 ./gstshark-plot ~/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50 [warning] Ill-shapen or missing "/* CTF x.y" header for text-only metadata. [warning] Unsupported CTF specification version 127.0. Trying anyway. [error] at line 1: token "": syntax error, unexpected $end

[error] Error creating AST [warning] Unable to open trace metadata for path "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50". [warning] [Context] Cannot open_trace of format ctf at path /home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50. [warning] [Context] cannot open trace "/home/tbuckley/tmp/seattle_tx2/VideoSystehark_2020-05-06_14:54:50 for reading. [error] Cannot open any trace for reading.

[error] opening trace "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-

[error] none of the specified trace paths could be opened.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

tmbuckley commented 4 years ago

this is on a nvidia jetston tx2, gstreamer is 1.14.x your gstshark had version of 0.6.x

files attached. datastream is a ctf file, metadata is empty.

Is there a complete 0.5.x tar file so I can go back a version, I found a 0.5.x.gz but it failed to git the common code in autogen.sh

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Wednesday, May 6, 2020 3:30 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

Can you share your trace files?

On May 6, 2020, at 14:13, tmbuckley notifications@github.com wrote:

? ./gstshark-plot ~/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50 [warning] Ill-shapen or missing "/* CTF x.y" header for text-only metadata. [warning] Unsupported CTF specification version 127.0. Trying anyway. [error] at line 1: token "": syntax error, unexpected $end

[error] Error creating AST [warning] Unable to open trace metadata for path "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50". [warning] [Context] Cannot open_trace of format ctf at path /home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50. [warning] [Context] cannot open trace "/home/tbuckley/tmp/seattle_tx2/VideoSystehark_2020-05-06_14:54:50 for reading. [error] Cannot open any trace for reading.

[error] opening trace "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-

[error] none of the specified trace paths could be opened.

- You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-624873654, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRX55DY3Q7QJLOLAS4LRQHCEZANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

?you have some debugging suggestions I can try for you, I really need this to work.

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Wednesday, May 6, 2020 3:30 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

Can you share your trace files?

On May 6, 2020, at 14:13, tmbuckley notifications@github.com wrote:

? ./gstshark-plot ~/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50 [warning] Ill-shapen or missing "/* CTF x.y" header for text-only metadata. [warning] Unsupported CTF specification version 127.0. Trying anyway. [error] at line 1: token "": syntax error, unexpected $end

[error] Error creating AST [warning] Unable to open trace metadata for path "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50". [warning] [Context] Cannot open_trace of format ctf at path /home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50. [warning] [Context] cannot open trace "/home/tbuckley/tmp/seattle_tx2/VideoSystehark_2020-05-06_14:54:50 for reading. [error] Cannot open any trace for reading.

[error] opening trace "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-

[error] none of the specified trace paths could be opened.

- You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-624873654, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRX55DY3Q7QJLOLAS4LRQHCEZANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

?since the datastream file is not understood, and the metadata is empty how does gstshark cause these files to be generated.

can I turn on debugging or look at libraries, or other things to help get this to work

Thanks,

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Wednesday, May 6, 2020 3:30 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

Can you share your trace files?

On May 6, 2020, at 14:13, tmbuckley notifications@github.com wrote:

? ./gstshark-plot ~/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50 [warning] Ill-shapen or missing "/* CTF x.y" header for text-only metadata. [warning] Unsupported CTF specification version 127.0. Trying anyway. [error] at line 1: token "": syntax error, unexpected $end

[error] Error creating AST [warning] Unable to open trace metadata for path "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50". [warning] [Context] Cannot open_trace of format ctf at path /home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50. [warning] [Context] cannot open trace "/home/tbuckley/tmp/seattle_tx2/VideoSystehark_2020-05-06_14:54:50 for reading. [error] Cannot open any trace for reading.

[error] opening trace "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-

[error] none of the specified trace paths could be opened.

- You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-624873654, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRX55DY3Q7QJLOLAS4LRQHCEZANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

found and compiled babeltrace2 and changed gstshark-plot to use babeltrace2

got this. at least it looks like babeltrace2 is getting farther with the datastream file.

So what am I missing, for tools and libs

Terry

05-07 16:54:53.935 20896 20896 W PLUGIN/SRC.CTF.FS/QUERY support_info_query@query.c:460 cannot append metadata content: metadata-decoder-status=-1 05-07 16:54:53.935 20896 20896 W LIB/QUERY-EXECUTOR bt_query_executor_query@query-executor.c:237 Component class's "query" method failed: query-exec-addr=0x559e878740, cc-addr=0x559e885020, cc-type=SOURCE, cc-name="fs", cc-partial-descr="Read CTF traces from the file sy", cc-is-frozen=0, cc-so-handle-addr=0x559e884480, cc-so-handle-path="/usr/local/lib/babeltrace2/plugins/babeltrace-plugin-ctf.la", object="babeltrace.support-info", params-addr=0x559e889230, params-type=MAP, params-element-count=2, log-level=WARNING 05-07 16:54:53.935 20896 20896 E CLI-CFG-SRC-AUTO-DISC support_info_query_all_sources@autodisc.c:493 babeltrace.support-info query failed. 05-07 16:54:53.935 20896 20896 E CLI main@babeltrace2.c:2645 Command-line error: retcode=1

ERROR: [Babeltrace CLI] (babeltrace2.c:2645) Command-line error: retcode=1 CAUSED BY [Source auto-discovery] (autodisc.c:493) babeltrace.support-info query failed. CAUSED BY [libbabeltrace2] (query-executor.c:237) Component class's "query" method failed: query-exec-addr=0x559e878740, cc-addr=0x559e885020, cc-type=SOURCE, cc-name="fs", cc-partial-descr="Read CTF traces from the file sy", cc-is-frozen=0, cc-so-handle-addr=0x559e884480, cc-so-handle-path="/usr/local/lib/babeltrace2/plugins/babeltrace-plugin-ctf.la", object="babeltrace.support-info", params-addr=0x559e889230, params-type=MAP, params-element-count=2, log-level=WARNING Loading proctime events... Loading interlatency events... Loading scheduling events... Loading framerate events... Loading bitrate events... Loading queuelevel events... Loading cpuusage events... Processing proctime... Processing interlatency... Processing framerate... Processing scheduling... Processing cpusage... Processing bitrate... Processing queuelevel...

?

From: Carl Michael Grüner Monzón notifications@github.com Sent: Wednesday, May 6, 2020 3:30 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

Can you share your trace files?

On May 6, 2020, at 14:13, tmbuckley notifications@github.com wrote:

? ./gstshark-plot ~/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50 [warning] Ill-shapen or missing "/* CTF x.y" header for text-only metadata. [warning] Unsupported CTF specification version 127.0. Trying anyway. [error] at line 1: token "": syntax error, unexpected $end

[error] Error creating AST [warning] Unable to open trace metadata for path "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50". [warning] [Context] Cannot open_trace of format ctf at path /home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-05-06_14:54:50. [warning] [Context] cannot open trace "/home/tbuckley/tmp/seattle_tx2/VideoSystehark_2020-05-06_14:54:50 for reading. [error] Cannot open any trace for reading.

[error] opening trace "/home/tbuckley/tmp/seattle_tx2/VideoSystem/gstshark_2020-

[error] none of the specified trace paths could be opened.

- You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-624873654, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRX55DY3Q7QJLOLAS4LRQHCEZANCNFSM4M2X2SAA.

michaelgruner commented 4 years ago

Looking at the trace files, one sec.

michaelgruner commented 4 years ago

I cannot find the attached trace files anywhere, I checked the email as well. Can you please send them over again?

On the other hand, please share the following:

tmbuckley commented 4 years ago

metadata is empty.

attached datastream.

My gst-launch is very complicated.

What are you looking for in the? gst_debug-2

thanks

terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 10:58 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I cannot find the attached trace files any where, I checked the email as well. Can you please send them over again?

On the other hand, please share the following:

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-625884994, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRUKUQGJYUA7OKGBVXTRQQT3RANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

​this has my command line also

this is the output from GST_DEBUG=3 running my system

Terry

Opened database successfully "main.cpp [ 51] main() :" prebuffer size is: 150 "main.cpp [ 62] main() :" stalkerTakePicMem 0x7fa8c56000 "main.cpp [ 63] main() :" stalkerCreateVideoMem 0x7fa8c55000 "main.cpp [ 64] main() :" stalkerVideoSystemMem 0x7fa8c54000 "mainwindow.cpp [ 16] MainWindow() :" this MainWindow(0x7fe767d850) "mainwindow.cpp [ 17] MainWindow() :" ui 0x55aef05a40 "mainwindow.cpp [ 18] MainWindow() :" instance MainWindow(0x7fe767d850) "videosystem.cpp [ 158] run() :" "videosystem.cpp [ 166] run() :" received data "videosystem.cpp [ 138] processBuf() :" CMDVIDEOSTART "gstinterface.cpp [ 137] GstInterface() :" GstInterface "gstinterface.cpp [ 152] InitGst() :" initGst mPipeline 0x0 "gstinterface.cpp [ 170] InitGst() :" initGst "gstinterface.cpp [ 172] InitGst() :" initGst message: len 0x32 0000 01 01 08 00 00 00 00 00 00 00 00 00 00 00 00 00 ................

InitGst(259) CMDLINE nvarguscamerasrc ! video/x-raw(memory:NVMM),width=3864,height=2180,format=NV12 ! nvvidconv ! video/x-raw,format=BGRx ! clockoverlay font-desc="calibri 20px" color=0xFF0000FF time-format="%D %H:%M:%S" halignment=right ! gdkpixbufoverlay name=logo location="/usr/local/stalker/Logo-500x3.svg" ! gdkpixbufoverlay name=retical location="/usr/local/stalker/crosshair-500x3.svg" ! textoverlay name=text font-desc="calibri 20px" color=0xFF0000FF valignment=top halignment=left xpad=0 ypad=100 text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue ! nveglglessink name=displaySink async=false sync=false t. ! queue ! stalkertakepicture ! nvjpegenc ! multifilesink name=stalkerPicture async=false location=/usr/local/stalker/tmp/Stalker_%05d.jpg t. ! queue ! nvvidconv ! video/x-raw(memory:NVMM),width=(int)1920,height=(int)1080,format=(string)I420 ! queue ! nvv4l2h264enc ! h264parse ! stalkercreatevideo name=stalkercreatevideo ! splitmuxsink name=stalkerVideoSink async-handling=true location=/tmp/tmpFile ../gst-plugin/src/gststalkertakepicture.c(163) CORRECT INIT PLACE 0x7fa8c53000 gst_stalkercreatevideo_init(159) InitGst(262) mPipeline 0x55af2f0170 InitGst(285) logo 0x55af294190 Opened database successfully Opened database successfully "gstinterface.cpp [ 40] getRetical() :" reticalType is: 0 "gstinterface.cpp [ 41] getRetical() :" reticleSize is: 500 InitGst(304) retical 0x55af2a1c30 Opened database successfully "gstinterface.cpp [ 703] UpdateText() :" textLocation is: 0 UpdateText(708) textOverlayImagel 0x55af286410 "gstinterface.cpp [ 741] UpdateText() :" No data in videoText InitGst(315) showSink 0x55af0f05c0

Using winsys: x11 0:00:00.200401983 17492 0x55aee95520 WARN   v4l2 gstv4l2object.c:2370:gst_v4l2_object_add_interlace_mode:0x55af2dfef0 Failed to determine interlace mode 0:00:00.200556386 17492 0x55aee95520 WARN   v4l2 gstv4l2object.c:2370:gst_v4l2_object_add_interlace_mode:0x55af2dfef0 Failed to determine interlace mode 0:00:00.200657443 17492 0x55aee95520 WARN   v4l2 gstv4l2object.c:2370:gst_v4l2_object_add_interlace_mode:0x55af2dfef0 Failed to determine interlace mode 0:00:00.200915591 17492 0x55aee95520 WARN   v4l2 gstv4l2object.c:4408:gst_v4l2_object_probe_caps: Failed to probe pixel aspect ratio with VIDIOC_CROPCAP: Unknown error -1 "mainwindow.cpp [ 44] VideoStart() :" gst.Init returned true 0:00:00.218374014 17492 0x55af3065e0 FIXME   default gstutils.c:3981:gst_pad_create_stream_id_internal: Creating random stream-id, consider implementing a deterministic way of creating a stream-id NvMMLiteOpen : Block : BlockType = 4 ===== NVMEDIA: NVENC ===== NvMMLiteBlockCreate : Block : BlockType = 4 0:00:00.236854309 17492 0x55af3068f0 WARN   v4l2bufferpool gstv4l2bufferpool.c:1062:gst_v4l2_buffer_pool_start: Uncertain or not enough buffers, enabling copy threshold H264: Profile = 66, Level = 0 0:00:01.269429015 17492 0x7f6c003ca0 WARN   v4l2bufferpool gstv4l2bufferpool.c:1526:gst_v4l2_buffer_pool_dqbuf: Driver should never set v4l2_buffer.field to ANY

​

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 10:58 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I cannot find the attached trace files any where, I checked the email as well. Can you please send them over again?

On the other hand, please share the following:

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-625884994, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRUKUQGJYUA7OKGBVXTRQQT3RANCNFSM4M2X2SAA.

michaelgruner commented 4 years ago

Mostly to see if tracers are printing any errors. Do simpler pipelines work? this can tell us if it's something in the environment or in the pipeline (a bug in gst-shark)

tmbuckley commented 4 years ago

?new debug output file,

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:11 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

Mostly to see if tracers are printing any errors. Do simpler pipelines work? this can tell us if it's something in the environment or in the pipeline (a bug in gst-shark)

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-625890728, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRQW55NLOHJUVEMRWN3RQQVJ7ANCNFSM4M2X2SAA.

michaelgruner commented 4 years ago

Please try this pipeline and see if the generated trace files are valid:

GST_TRACERS=proctime gst-launch-1.0 videotestsrc ! identity ! fakesink sync=true

michaelgruner commented 4 years ago

FYI, if you are attaching files, I cannot see them anywhere.

tmbuckley commented 4 years ago

?need a better way to share files.

Need to take a break , will try to get more data for you in an hour.

gst-shark-plot did not throw any errors on the videotestsrc line you sent, but there was no plot either.

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:24 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

FYI, if you are attaching files, I cannot see them anywhere.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-625896591, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRQ6GQGZKXXFZAGDFUTRQQW47ANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

?so it looks like your test line works, and my multi "t" multi-thread and multi-processor environment caused the datastream file to be corrupted.

too many processes writing to the datastream and missing syncs, for flushes?

Can I try anything else to get you more info?

Thanks,

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:24 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

FYI, if you are attaching files, I cannot see them anywhere.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-625896591, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRQ6GQGZKXXFZAGDFUTRQQW47ANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

?my brief looking at the cft.c I think I remember parts being written without any locks between cores/threads/processes.

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:24 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

FYI, if you are attaching files, I cannot see them anywhere.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-625896591, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRQ6GQGZKXXFZAGDFUTRQQW47ANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

GST_TRACERS=cpuusage gst-launch-1.0 videotestsrc ! identity ! fakesink sync=true?

did not generate a datastream or metadata file.

Thanks,

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:24 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

FYI, if you are attaching files, I cannot see them anywhere.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-625896591, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRQ6GQGZKXXFZAGDFUTRQQW47ANCNFSM4M2X2SAA.

michaelgruner commented 4 years ago

Ah, good catch. Let me check.

tmbuckley commented 4 years ago

?thank you,

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 1:30 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

Ah, good catch. Let me check.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-625954308, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRTAXIRJSPBFGZJELKTRQRFTZANCNFSM4M2X2SAA.

michaelgruner commented 4 years ago

I tested a somewhat similar pipe to yours in my laptop (see below). I didn't get the same problem, but I did get a segfault when using the interlatency plugin. The cause was an unlinked pad in the pipe which we didn't handle.

text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! ****queue**** t. ! queue !

I'm thinking this could be the reason your trace files were corrupted. Can you try the following branch hotfix/interlatency-segfault-on-unlinked-pads and see if you get a different result?

Here's the full pipe based on yours:

GST_TRACERS="proctime;scheduletime;interlatency;framerate;queuelevel;bitrate;buffer" gst-launch-1.0 avfvideosrc ! video/x-raw,width=1280,height=720,format=NV12 ! videoconvert ! video/x-raw,format=BGRx ! clockoverlay font-desc="calibri 20px" color=0xFF0000FF time-format="%D %H:%M:%S" halignment=right ! gdkpixbufoverlay name=logo  location="crosshair.svg" ! gdkpixbufoverlay name=retical location="crosshair.svg" ! textoverlay name=text font-desc="calibri 20px" color=0xFF0000FF  valignment=top halignment=left xpad=0 ypad=100 text="50.5 MPH  300.0 FT" ! queue  ! tee name=t ! queue  t. ! queue     ! glimagesink async=false sync=false t. ! queue     ! jpegenc ! multifilesink name=stalkerPicture async=false location=/tmp/Stalker.jpg t. ! queue  ! videoconvert ! videoscale ! video/x-raw,width=1920,height=1080,format=I420    ! queue     ! x264enc  ! h264parse    ! splitmuxsink name=stalkerVideoSink async-handling=true location=/tmp/tmpFile
tmbuckley commented 4 years ago

did a git pull origin hotfix/interlatency-segfault-on-unlinked-pads?

make; sudo make install

my test did not run, it just hangs, added gdb and nothing is crashing, only no video.

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:22 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I tested a somewhat similar pipe to yours in my laptop (see below). I didn't get the same problem, but I did get a segfault when using the interlatency plugin. The cause was an unlinked pipe in the pipe which we didn't handle.

text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue !

I'm thinking this could be the reason your trace files were corrupted. Can you try the following branch hotfix/interlatency-segfault-on-unlinked-pads and see if you get a different result?

Here's the full pipe based on yours:

GST_TRACERS="proctime;scheduletime;interlatency;framerate;queuelevel;bitrate;buffer" gst-launch-1.0 avfvideosrc ! video/x-raw,width=1280,height=720,format=NV12 ! videoconvert ! video/x-raw,format=BGRx ! clockoverlay font-desc="calibri 20px" color=0xFF0000FF time-format="%D %H:%M:%S" halignment=right ! gdkpixbufoverlay name=logo location="crosshair.svg" ! gdkpixbufoverlay name=retical location="crosshair.svg" ! textoverlay name=text font-desc="calibri 20px" color=0xFF0000FF valignment=top halignment=left xpad=0 ypad=100 text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue ! glimagesink async=false sync=false t. ! queue ! jpegenc ! multifilesink name=stalkerPicture async=false location=/tmp/Stalker.jpg t. ! queue ! videoconvert ! videoscale ! video/x-raw,width=1920,height=1080,format=I420 ! queue ! x264enc ! h264parse ! splitmuxsink name=stalkerVideoSink async-handling=true location=/tmp/tmpFile

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626103396, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRURZLUTCZUPJJHZ4BDRQTK7NANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

?debugging pointers? Tried gdb, nothing died, but my stream is hung

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:22 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I tested a somewhat similar pipe to yours in my laptop (see below). I didn't get the same problem, but I did get a segfault when using the interlatency plugin. The cause was an unlinked pipe in the pipe which we didn't handle.

text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue !

I'm thinking this could be the reason your trace files were corrupted. Can you try the following branch hotfix/interlatency-segfault-on-unlinked-pads and see if you get a different result?

Here's the full pipe based on yours:

GST_TRACERS="proctime;scheduletime;interlatency;framerate;queuelevel;bitrate;buffer" gst-launch-1.0 avfvideosrc ! video/x-raw,width=1280,height=720,format=NV12 ! videoconvert ! video/x-raw,format=BGRx ! clockoverlay font-desc="calibri 20px" color=0xFF0000FF time-format="%D %H:%M:%S" halignment=right ! gdkpixbufoverlay name=logo location="crosshair.svg" ! gdkpixbufoverlay name=retical location="crosshair.svg" ! textoverlay name=text font-desc="calibri 20px" color=0xFF0000FF valignment=top halignment=left xpad=0 ypad=100 text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue ! glimagesink async=false sync=false t. ! queue ! jpegenc ! multifilesink name=stalkerPicture async=false location=/tmp/Stalker.jpg t. ! queue ! videoconvert ! videoscale ! video/x-raw,width=1920,height=1080,format=I420 ! queue ! x264enc ! h264parse ! splitmuxsink name=stalkerVideoSink async-handling=true location=/tmp/tmpFile

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626103396, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRURZLUTCZUPJJHZ4BDRQTK7NANCNFSM4M2X2SAA.

michaelgruner commented 4 years ago

I don’t think the tracers can hang your stream. Can you look for anything off when running with GST_DEBUG=2? I’m very suspicious of that unlinked pad you have on your pipeline.

On May 11, 2020, at 10:08, tmbuckley notifications@github.com wrote:

 ?debugging pointers? Tried gdb, nothing died, but my stream is hung

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:22 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I tested a somewhat similar pipe to yours in my laptop (see below). I didn't get the same problem, but I did get a segfault when using the interlatency plugin. The cause was an unlinked pipe in the pipe which we didn't handle.

text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue !

I'm thinking this could be the reason your trace files were corrupted. Can you try the following branch hotfix/interlatency-segfault-on-unlinked-pads and see if you get a different result?

Here's the full pipe based on yours:

GST_TRACERS="proctime;scheduletime;interlatency;framerate;queuelevel;bitrate;buffer" gst-launch-1.0 avfvideosrc ! video/x-raw,width=1280,height=720,format=NV12 ! videoconvert ! video/x-raw,format=BGRx ! clockoverlay font-desc="calibri 20px" color=0xFF0000FF time-format="%D %H:%M:%S" halignment=right ! gdkpixbufoverlay name=logo location="crosshair.svg" ! gdkpixbufoverlay name=retical location="crosshair.svg" ! textoverlay name=text font-desc="calibri 20px" color=0xFF0000FF valignment=top halignment=left xpad=0 ypad=100 text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue ! glimagesink async=false sync=false t. ! queue ! jpegenc ! multifilesink name=stalkerPicture async=false location=/tmp/Stalker.jpg t. ! queue ! videoconvert ! videoscale ! video/x-raw,width=1920,height=1080,format=I420 ! queue ! x264enc ! h264parse ! splitmuxsink name=stalkerVideoSink async-handling=true location=/tmp/tmpFile

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626103396, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRURZLUTCZUPJJHZ4BDRQTK7NANCNFSM4M2X2SAA. — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

tmbuckley commented 4 years ago

?add the tracer stuff and my app hangs, remove tracer stuff app runs and displays video.

Please help me debug this, really need to determine cpuusage for cleanup.

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Monday, May 11, 2020 11:13 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I don't think the tracers can hang your stream. Can you look for anything off when running with GST_DEBUG=2? I'm very suspicious of that unlinked pad you have on your pipeline.

On May 11, 2020, at 10:08, tmbuckley notifications@github.com wrote:

? ?debugging pointers? Tried gdb, nothing died, but my stream is hung

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:22 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I tested a somewhat similar pipe to yours in my laptop (see below). I didn't get the same problem, but I did get a segfault when using the interlatency plugin. The cause was an unlinked pipe in the pipe which we didn't handle.

text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue !

I'm thinking this could be the reason your trace files were corrupted. Can you try the following branch hotfix/interlatency-segfault-on-unlinked-pads and see if you get a different result?

Here's the full pipe based on yours:

GST_TRACERS="proctime;scheduletime;interlatency;framerate;queuelevel;bitrate;buffer" gst-launch-1.0 avfvideosrc ! video/x-raw,width=1280,height=720,format=NV12 ! videoconvert ! video/x-raw,format=BGRx ! clockoverlay font-desc="calibri 20px" color=0xFF0000FF time-format="%D %H:%M:%S" halignment=right ! gdkpixbufoverlay name=logo location="crosshair.svg" ! gdkpixbufoverlay name=retical location="crosshair.svg" ! textoverlay name=text font-desc="calibri 20px" color=0xFF0000FF valignment=top halignment=left xpad=0 ypad=100 text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue ! glimagesink async=false sync=false t. ! queue ! jpegenc ! multifilesink name=stalkerPicture async=false location=/tmp/Stalker.jpg t. ! queue ! videoconvert ! videoscale ! video/x-raw,width=1920,height=1080,format=I420 ! queue ! x264enc ! h264parse ! splitmuxsink name=stalkerVideoSink async-handling=true location=/tmp/tmpFile

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626103396, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRURZLUTCZUPJJHZ4BDRQTK7NANCNFSM4M2X2SAA.

You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626801472, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRW7MNPOS2T5HI6KN7LRRAP3JANCNFSM4M2X2SAA.

michaelgruner commented 4 years ago

Let me have someone try your pipe on a Jetson.

On May 11, 2020, at 10:18 AM, tmbuckley notifications@github.com wrote:

?add the tracer stuff and my app hangs, remove tracer stuff app runs and displays video.

Please help me debug this, really need to determine cpuusage for cleanup.

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Monday, May 11, 2020 11:13 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I don't think the tracers can hang your stream. Can you look for anything off when running with GST_DEBUG=2? I'm very suspicious of that unlinked pad you have on your pipeline.

On May 11, 2020, at 10:08, tmbuckley notifications@github.com wrote:

? ?debugging pointers? Tried gdb, nothing died, but my stream is hung

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:22 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I tested a somewhat similar pipe to yours in my laptop (see below). I didn't get the same problem, but I did get a segfault when using the interlatency plugin. The cause was an unlinked pipe in the pipe which we didn't handle.

text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue !

I'm thinking this could be the reason your trace files were corrupted. Can you try the following branch hotfix/interlatency-segfault-on-unlinked-pads and see if you get a different result?

Here's the full pipe based on yours:

GST_TRACERS="proctime;scheduletime;interlatency;framerate;queuelevel;bitrate;buffer" gst-launch-1.0 avfvideosrc ! video/x-raw,width=1280,height=720,format=NV12 ! videoconvert ! video/x-raw,format=BGRx ! clockoverlay font-desc="calibri 20px" color=0xFF0000FF time-format="%D %H:%M:%S" halignment=right ! gdkpixbufoverlay name=logo location="crosshair.svg" ! gdkpixbufoverlay name=retical location="crosshair.svg" ! textoverlay name=text font-desc="calibri 20px" color=0xFF0000FF valignment=top halignment=left xpad=0 ypad=100 text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue ! glimagesink async=false sync=false t. ! queue ! jpegenc ! multifilesink name=stalkerPicture async=false location=/tmp/Stalker.jpg t. ! queue ! videoconvert ! videoscale ! video/x-raw,width=1920,height=1080,format=I420 ! queue ! x264enc ! h264parse ! splitmuxsink name=stalkerVideoSink async-handling=true location=/tmp/tmpFile

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626103396, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRURZLUTCZUPJJHZ4BDRQTK7NANCNFSM4M2X2SAA.

You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626801472, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRW7MNPOS2T5HI6KN7LRRAP3JANCNFSM4M2X2SAA. — You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626804245, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAFPIOHHES73QSP4633UGH3RRAQOHANCNFSM4M2X2SAA.

tmbuckley commented 4 years ago

​Is there a place or FAQ on debugging gstshark?

I can do things to figure out what is happening, how do I get info out of the stream to figure out why gstshark is killing my gstreamer app.

Thanks,

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Monday, May 11, 2020 11:22 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

Let me have someone try your pipe on a Jetson.

On May 11, 2020, at 10:18 AM, tmbuckley notifications@github.com wrote:

?add the tracer stuff and my app hangs, remove tracer stuff app runs and displays video.

Please help me debug this, really need to determine cpuusage for cleanup.

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Monday, May 11, 2020 11:13 AM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I don't think the tracers can hang your stream. Can you look for anything off when running with GST_DEBUG=2? I'm very suspicious of that unlinked pad you have on your pipeline.

On May 11, 2020, at 10:08, tmbuckley notifications@github.com wrote:

? ?debugging pointers? Tried gdb, nothing died, but my stream is hung

Terry

From: Carl Michael Grüner Monzón notifications@github.com Sent: Friday, May 8, 2020 11:22 PM To: RidgeRun/gst-shark Cc: Terry Buckley; Author Subject: Re: [RidgeRun/gst-shark] gstshark not working, datastream and metadata bad? (#69)

I tested a somewhat similar pipe to yours in my laptop (see below). I didn't get the same problem, but I did get a segfault when using the interlatency plugin. The cause was an unlinked pipe in the pipe which we didn't handle.

text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue !

I'm thinking this could be the reason your trace files were corrupted. Can you try the following branch hotfix/interlatency-segfault-on-unlinked-pads and see if you get a different result?

Here's the full pipe based on yours:

GST_TRACERS="proctime;scheduletime;interlatency;framerate;queuelevel;bitrate;buffer" gst-launch-1.0 avfvideosrc ! video/x-raw,width=1280,height=720,format=NV12 ! videoconvert ! video/x-raw,format=BGRx ! clockoverlay font-desc="calibri 20px" color=0xFF0000FF time-format="%D %H:%M:%S" halignment=right ! gdkpixbufoverlay name=logo location="crosshair.svg" ! gdkpixbufoverlay name=retical location="crosshair.svg" ! textoverlay name=text font-desc="calibri 20px" color=0xFF0000FF valignment=top halignment=left xpad=0 ypad=100 text="50.5 MPH 300.0 FT" ! queue ! tee name=t ! queue t. ! queue ! glimagesink async=false sync=false t. ! queue ! jpegenc ! multifilesink name=stalkerPicture async=false location=/tmp/Stalker.jpg t. ! queue ! videoconvert ! videoscale ! video/x-raw,width=1920,height=1080,format=I420 ! queue ! x264enc ! h264parse ! splitmuxsink name=stalkerVideoSink async-handling=true location=/tmp/tmpFile

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626103396, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRURZLUTCZUPJJHZ4BDRQTK7NANCNFSM4M2X2SAA.

You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.

- You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626801472, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRW7MNPOS2T5HI6KN7LRRAP3JANCNFSM4M2X2SAA. — You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626804245, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAFPIOHHES73QSP4633UGH3RRAQOHANCNFSM4M2X2SAA.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/RidgeRun/gst-shark/issues/69#issuecomment-626806610, or unsubscribehttps://github.com/notifications/unsubscribe-auth/APPLNRQGFZ2PMQPTEVT4A73RRAQ5JANCNFSM4M2X2SAA.

michaelgruner commented 4 years ago

So I'm unable to reproduce your problem using a similar pipe to yours on my PC. @maumonteroj is reproducing your pipe on a TX2. @maumonteroj do you have any updates?

michaelgruner commented 4 years ago

@tmbuckley does this happen with any tracer or just CPU usage?

maumonteroj commented 4 years ago

Hi @tmbuckley, I was able to reproduce your issue in my tx2, we are going to continue checking how to solve it.

maumonteroj commented 4 years ago

Hi @tmbuckley , we found that the issue is because your pipeline is not closing properly. In order to close I need to force the teardown and for that reason, the pending file flushes are not made it and the files get corrupted or empty.

I made a branch for you to test in order to see if this is the same issue you are experimenting. Can you please try the test/disable-file-buffering branch and see if you get a different result?

We are going to evaluate make this solution a configurable variable.

alehed commented 4 years ago

Hey @maumonteroj, thanks a lot for the branch! I was having the same issue. Compiling gst-shark from the branch fixed it for me.

compiaffe commented 4 years ago

@maumonteroj @michaelgruner I also had that issue. Using the test/disable-file-buffering branch fixed it for me, too.

michaelgruner commented 4 years ago

Thanks all for the confirmation, I'll close this issue. We're making a release later this week with a few improvements, including this one.