Bug 856 - 99% CPU
: 99% CPU
Status: RESOLVED FIXED
Product: Logitech Media Server
Classification: Unclassified
Component: Audio
: unspecified
: PC Windows XP
: P2 normal (vote)
: ---
Assigned To: Dan Sully
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2005-02-24 13:59 UTC by Bart
Modified: 2008-09-15 14:37 UTC (History)
0 users

See Also:
Category: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Bart 2005-02-24 13:59:37 UTC
How much of a glitch in playback is acceptable?  I would suggest that 
practically no glitch is acceptable.  My CD player has never skipped in 12 
years, my Winamp system only skips a few times a year (and I listen to music a 
lot - several hours a day).  Slimserver locks up solid every day and not only 
stops playing music but also eats 99% of my CPU and locks my PC.  Today it 
locked up for 5.5 mins after it imported data from MusicMagic, later (as we 
were listening to music) it locked up for 16 minutes after it re-imported data 
from MusicMagic.
Will this ever settle down to no hangups, transparent-to-non-computer-literate-
people useability?
In the hope of getting Slimserver and MusicMagic working I have built a 
dedicated, fairly well specced PC, but I was hoping (and am hoping in the 
future) to be able to run it on a spare low-spec machine.  Am I dreaming?
Comment 1 KDF 2005-02-25 01:08:15 UTC
you arent dreaming, but this is an alpha. Musicmagic is also an API that relies
on http transfers for information.  Windows makes it difficult to do an http
transaction while allowing other activities to progress, espectially if the
transaction has some holdup.  Grabbing the genre list and artist list are large
transactions and have the potential of problems, tho I dont current see them on
my  setup (only 400 songs on windows, however).  It is a known problem, and it
is being worked on.

What could help in this effort would be if you can confirm the timing of these
dropouts.  run the server with --d_musicmagic and --d_info.  you can attach the
sections of the log that match up with your dropouts to this bug report.

for the occastion when musicmagic isn't running, Iv'e shortened the timeout so
that the attempts by the server to check for musicmagic don't cause dropouts, or
at least reduce them.  revision 2215.
Comment 2 Bart 2005-02-25 01:33:19 UTC
No, I know I am not dreaming, I am generally very happy with this hardware and 
software. Yesterday I was just a bit put out that it dropped for 16 minutes in 
the evening when we were listening to it, and I had just been saying how good 
it was!
Are these the relevant sections of the log? (I was only using --d_musicmagic, 
but I will run it today with --d_info also)
2005-02-24 13:28:55.5859 MusicMagic: Got 145 active genre(s).
2005-02-24 13:29:03.6661 MusicMagic: Got 4365 active artist(s).
2005-02-24 13:34:33.1814 MusicMagic: Found playlist: musicmagicplaylist:Watch%
20List
2005-02-24 13:34:33.2095 MusicMagic: done Scanning

2005-02-24 20:08:06.4180 MusicMagic: Got 145 active genre(s).
2005-02-24 20:08:14.0255 MusicMagic: Got 4365 active artist(s).
getpeername() on closed socket GEN10 at C:/Perl/lib/IO/Socket.pm line 206.
2005-02-24 20:24:07.4980 MusicMagic: done Scanning
2005-02-24 20:24:07.7339 exportFunction: finished export (0 records, 1 
playlists)
Comment 3 Bart 2005-02-25 02:17:35 UTC
OK, here is a section of the log from today, running with --d_musicmagic and --
d_info, just come back from a 21 minute lockup.
2005-02-25 09:45:50.9543 Updating file:///M:/Impressions,%20The/Impressions%20-
%20Keep%20On%20Pushin'.mp3 : BITRATE to 256
Malformed UTF-8 character (unexpected end of string) at C:/Program 
Files/SlimServer/server/Slim/Player/Player.pm line 245.
2005-02-25 09:45:51.1209 mp3 file type for http://localhost:10002/api/genres?
active
2005-02-25 09:45:51.1254 MusicMagic: Got 145 active genre(s).
Malformed UTF-8 character (unexpected end of string) at C:/Program 
Files/SlimServer/server/Slim/Player/Player.pm line 245.
2005-02-25 09:45:58.8341 mp3 file type for http://localhost:10002/api/artists?
active
2005-02-25 09:45:58.8658 MusicMagic: Got 4365 active artist(s).
Malformed UTF-8 character (unexpected end of string) at C:/Program 
Files/SlimServer/server/Slim/Player/Player.pm line 245.
2005-02-25 10:07:11.2966 mp3 file type for http://localhost:10002/api/playlists
2005-02-25 10:07:11.2988 MusicMagic: done Scanning
2005-02-25 10:07:11.3099 mp3 file type for http://localhost:10002/api/cacheid
2005-02-25 10:07:11.4334 Finished background scanning.
2005-02-25 10:07:11.4625 forceCommit: syncing to the database.
2005-02-25 10:07:11.8563 exportFunction: finished export (0 records, 1 
playlists)
2005-02-25 10:07:11.8666 DBI: Supressing periodic commit - no dirty items
Comment 4 KDF 2005-02-25 09:38:29 UTC
Those logs are meaningless unless you tell us when in those logs that you
experience a dropout in audio.  There are a few suspects, but confirmation is
needed to be sure that the fix covers all the cases that you experience.
Comment 5 Bart 2005-02-25 10:26:09 UTC
The dropouts occurred at 13:29, returning at 13:34, again at 20:08, returning 
at 20:24 (by which time I had fired up the old Winamp system...) and today at 
09:45, returning at 10:07
I can post more of the log, or send you the whole thing (6Mb for yesterday, 
36Mb so far today)

Comment 6 Dan Sully 2005-02-25 11:15:40 UTC
Bart - I believe I've isolated the problem code. I'll try to get in a fix today.
Longer term, we need to have a change from Predixis to help optimize the
integration.
Comment 7 Dan Sully 2005-02-25 15:57:37 UTC
Bart - I just checked in subversion change #2225 - which should fix the problem
you are seeing here. Can you please confirm?

Thanks. 
Comment 8 Bart 2005-02-26 04:31:22 UTC
Didn't get to check subversion change #2225 as we went out to see Hank 
Wangford last night, but I have installed today's build and imported the 
MusicMagic data with no noticable interruption to the playback.
Comment 9 Dan Sully 2005-02-26 10:19:50 UTC
Great! I'm going to close this bug.
Comment 10 Chris Owens 2006-06-16 14:40:55 UTC
There are 536 bugs in the database with targets of '---' that were fixed prior to new year 2006.  I am setting them to targets of 6.2.1 to keep them from showing up in my queries.