View unanswered posts    View active topics

All times are UTC - 6 hours





Post new topic Reply to topic  [ 19 posts ] 
Go to page Previous  1, 2

Print view Previous topic   Next topic  
Author Message
Search for:
PostPosted: Mon Jul 04, 2016 10:58 am 
Offline
Joined: Fri Nov 19, 2004 11:58 pm
Posts: 99
brfransen wrote:
sudo sv restart frontend is probably the most "proper" way. And we have that tied to the Alt+f shortcut. But if that is not working that it would seem to indicate that Xorg is what is wedged and not just mythfrontend. When it freezes are you able to open a xterm or ssh in?.


SSH into the box is always available.
From the frontend a Ctrl-X does start an xterm, there is a a few second delay on launch as to when the cat walks across the keyboard and hits some key combination with enter key and successfully launches a few hundred xterms in under two seconds.

[quoteAnother note is that one change in R8.4 was that when the frontend shuts down now it tries to kill all mythfrontend explicitly. Before it relied on Xorg taking them down but I have seen some situations where the mythfrontend process didn't quit when Xorg went away.[/quote]

My new hobby of watching htop, shows the some unamed process is temporarily able to gain privileged access to the cpu after the mythfrontend processes go 100% critical. I believe the PPID or PPGRP is 5 of the mystery process with a life of about 5 seconds. After this process goes away, within another 5 seconds the mythfrontend 100% behavior returns.


Top
 Profile  
 
PostPosted: Mon Jul 04, 2016 11:08 am 
Offline
Joined: Fri Nov 19, 2004 11:58 pm
Posts: 99
[quote="thekingofspain"][quote="brfransen"]Next time it happens can you see what other processes are zombies?
Code:
root      1064  0.0  0.0      0     0 ?        Z    23:14   0:00 [funcd] <defunct>
root      1333  0.0  0.0  12576  2260 pts/1    S+   23:15   0:00 grep defunct
mythtv   18815  5.6  0.0      0     0 ?        Zl   04:31  63:28 [mythfrontend] <defunct>


The defunct is similar with the latest pacman drop as before:

Code:
mythtv    3392 10.2  0.0      0     0 ?        Zl   02:12  60:27 [mythfrontend] <defunct>
root     17555  0.0  0.0      0     0 ?        Z    12:00   0:00 [funcd] <defunct>
root     17685  0.0  0.0  12576  2224 pts/1    S+   12:00   0:00 grep defunct


Also note the sv restart does kill the old mythfrontend process and starts a new one but the new process jumps to 100% like the old one. The main terminal is dead once once the mythfrontend goes 100%, and only ssh is available.

When attempting to reboot in this state, the current process is the following:
Code:
shutdown -r &
sv stop stop mythbackend
sv stop mysql
pkill -9 xinit
pkill -9 Xorg

The background shutdown is because the foreground shutdown never returns. However the at this point the background shutdown -r command is still being blocked. Killing the following process constantly allows the background shutdown process to trigger:
Code:
pkill -9 chvt


Top
 Profile  
 
PostPosted: Sat Jul 09, 2016 8:33 pm 
Offline
Joined: Fri Nov 19, 2004 11:58 pm
Posts: 99
Ok I have figured out a way to restart the frontend after a video playback with 100% success.

sv stop frontend

this makes mythfrontend go to 100%

htop

wait for the mythfrontend process to go from 100% to termination on its own. it takes about 90 seconds

sv start frontend

My guess why restart and manual performing a stop/start fail is because the mythfrontend never fully shutdowns and then a new mythfrontend starts up and attaches to the same process or device that original mythfrontend is waiting to be terminated, causing the race condition.


Top
 Profile  
 
PostPosted: Sat Sep 17, 2016 2:02 am 
Offline
Joined: Fri Nov 19, 2004 11:58 pm
Posts: 99
Each drop of LinHes is changing the pattern of how to recover from a video freeze without running into 100% CPU usage.

I am using the following command to recover is good success:

Code:
pkill -9 -o -f "su - mythtv -l -c /usr/LH/bin/LinHES-start"


Note that the CPU still jumps to 100% on an inner zombie process still gets trigger by the above kill command. However after about 20 seconds when the sv process automatically kicks off a new frontend, sv or X somehow is able to kill the 100% zombie process. The zombie process seem to be owned by the LinHES-start process and has no process name, however maybe the process is actually owned by the sv process or transferred to it.

The previous drop of LinHES, the 100% process in the zombie mythfrontend process was called AudioBaseOutput from recollection. Killing that inner process on a video freeze would free the video freeze without restarting the frontend entirely. The new drop, it not displaying any inner AudioBaseOutpt process on video hang.


Top
 Profile  
 

Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 19 posts ] 
Go to page Previous  1, 2



All times are UTC - 6 hours




Who is online

Users browsing this forum: No registered users and 7 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  
Powered by phpBB® Forum Software © phpBB Group

Theme Created By ceyhansuyu