Bug #2793
closed
EPGGRAP hanging
Added by Rafal Kupiec about 10 years ago. Updated almost 10 years ago.
0%
Description
I have updated TVH and after restart it tried to grab EPG OTA. This is still shown in subscriptions list and until restart its not updating EPG.
Screenshow was done today (20-04-2015) and it shows epggrab threads running for 2 days already.
Files
| tvh-epggrab.png (10.2 KB) tvh-epggrab.png | Rafal Kupiec, 2015-04-20 20:50 | ||
| debug.log (187 KB) debug.log | Thomas Knoll, 2015-05-07 13:46 | ||
| epggrab.png (14.1 KB) epggrab.png | Thomas Knoll, 2015-05-07 18:50 | ||
| epgsettings.png (46.2 KB) epgsettings.png | Thomas Knoll, 2015-05-07 18:56 |
Updated by Jaroslav Kysela almost 10 years ago
Could you retest with latest ? v3.9-2719-ga1844b8
Updated by Jaroslav Kysela almost 10 years ago
I cannot reproduce this here. Do you have some logs ? If the subcriptions stalls in the "testing", something is bad, because the timeout should catch them.
Updated by Thomas Knoll almost 10 years ago
I have the same problem here on 3.9.2765. I recognized it a few weeks ago as my powersaving method checks the state of active subscriptions. And right now I do have two "testing" epggrab processes again.
I will restart the server and try to catch a debug log.
Updated by Thomas Knoll almost 10 years ago
OK, so here is the logfile with a trace of epggrab. Hope you can find out whats wrong.
The server is running for two hours now, two epggrab processes running from the beginning with state "testing".
Updated by Jaroslav Kysela almost 10 years ago
Which subscriptions (which muxes) were in the failing state ?
Updated by Thomas Knoll almost 10 years ago
Should there have been any? Don't remember having any failed state subs/muxes and right now I don't see any. When epggrab is stalled in "testing" there is also no entry in the "service" column.
Interesting thing is, that is doesn't seem to be everytime epg scans are running. A few minutes ago I restarted tvheadend and the processes for epggrab (got two sundteck dvb-c) were all fine and after 5 minutes gone from the active subscription page. Another restart and the problem is there again, one process hanging with "testing" and no service entry.
What can I do to provide more information for debug?
Updated by Thomas Knoll almost 10 years ago
Maybe of interest: I restarted the tvheadend processes again and again... when the epggrab process is stalled it is always right at the beginning of a new started tvheadend. Then I got an epggrab while tvheadend is running its scan processes. This epggrab process will apparently hang and won't be killed. I tried it for about 20 times now, always the same... when everything is fine, tvheadend is starting its scan processes and epggrab is running after the scan. When it hangs, its starting both right from the beginning.
Updated by Thomas Knoll almost 10 years ago
- File epggrab.png epggrab.png added
This is what then happens. Right here I started tvheadend, it then was running two scan processes (two dvb-c cards) and two stalled epggrab processes. After the scan it is running the "normal" epggrab processes which will be stopped fined, the two remaining epggrab (the ones without service entry) will be there forever.
Updated by Thomas Knoll almost 10 years ago
- File epgsettings.png epgsettings.png added
Just for completion... my epg grabber settings.
Updated by Jaroslav Kysela almost 10 years ago
Do you see any 'service instance is bad, reason:' warnings in the log ? I assume that you're using the latest tvh code.
Updated by Thomas Knoll almost 10 years ago
negative. not in the normal log. do I have to trace for this?
Updated by B C almost 10 years ago
does this also happen if you disable initial scan and idle scanning?
Updated by Jaroslav Kysela almost 10 years ago
I added to the recent tvh code the 'Id' field in the subscription tab which can help to identify the stale point using the log. Could you reproduce the issue and enable the trace log using --trace subscription,service,mpegts ? Also, report back the stalled subscriptions in the webui.
Updated by Rafal Kupiec almost 10 years ago
Jaroslav Kysela wrote:
I added to the recent tvh code the 'Id' field in the subscription tab which can help to identify the stale point using the log. Could you reproduce the issue and enable the trace log using --trace subscription,service,mpegts ? Also, report back the stalled subscriptions in the webui.
This is what dmesg says about this problem:
[8183667.476214] usb 2-1: DVB: adapter 4 frontend 0 frequency 0 out of range (174000000..862000000)
Updated by Jaroslav Kysela almost 10 years ago
Rafal Kupiec wrote:
Jaroslav Kysela wrote:
I added to the recent tvh code the 'Id' field in the subscription tab which can help to identify the stale point using the log. Could you reproduce the issue and enable the trace log using --trace subscription,service,mpegts ? Also, report back the stalled subscriptions in the webui.
This is what dmesg says about this problem:
[8183667.476214] usb 2-1: DVB: adapter 4 frontend 0 frequency 0 out of range (174000000..862000000)
Irrelevant.
Updated by Jaroslav Kysela almost 10 years ago
- Status changed from New to Fixed
It should be fixed in v4.1-118-g81c356c and (v4.0.5 - when released).