Bug 608 - Scan 100% processor
: Scan 100% processor
Status: RESOLVED FIXED
Product: Logitech Media Server
Classification: Unclassified
Component: Windows Service
: 5.x or older
: PC Windows XP
: P2 normal (vote)
: ---
Assigned To: Blackketter Dean
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2004-10-05 02:24 UTC by Stuart Beesley
Modified: 2008-12-18 11:51 UTC (History)
1 user (show)

See Also:
Category: ---


Attachments
Log showing where scan stops on latest builds (52.80 KB, text/plain)
2004-10-05 02:25 UTC, Stuart Beesley
Details
Log showing that scan is OK on builds 2004-10-02 and earlier (1.88 MB, text/plain)
2004-10-05 02:28 UTC, Stuart Beesley
Details
This file on its own causes 100% processor (1.97 MB, audio/x-ms-wma)
2004-10-07 15:04 UTC, Stuart Beesley
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Stuart Beesley 2004-10-05 02:24:26 UTC
Windows XP. When I use builds 2004-10-03 or 2004-10-04 and I wipe the cache and 
do a rebuild the processor jumps to 100% and even after 2 hours the scan will 
not complete. Whilst it is scanning, after a few seconds the .db file stops 
growing, the hard disk stops churning. If I use --d_info and --d_scan, I can 
see the log file stops growing after a few seconds.

On build 2004-10-02 and all prior versions this does not happen. If I wipe the 
cache (or delete the .db)and start a rebuild, the processor stays at 30% max 
and the rebuild is complete after 3-5 mins.

I have 5000 songs.

I will attach two logs which show where the scan stops - even though the 
processor remains at 100%. The other on the prior build showing that it gets 
past this point. I can see no differences in the log.
Comment 1 Stuart Beesley 2004-10-05 02:25:07 UTC
Created attachment 169 [details]
Log showing where scan stops on latest builds
Comment 2 Stuart Beesley 2004-10-05 02:28:15 UTC
Created attachment 170 [details]
Log showing that scan is OK on builds 2004-10-02 and earlier

I stopped this after a minute or so, but it shows that the scan continues past
the point where the lastest build scan stops
Comment 3 KDF 2004-10-05 11:03:40 UTC
if you temporarily remove all the songs from ABBA Gold from your library, does
it scan?  each time it stops, try removing that one song from the library
temporarily until it gets through.  then you'll have a folder full of files that
it gets stuck on.  One of those files may be useful to Dean to test this problem.
Comment 4 Chris 2004-10-07 02:02:39 UTC
I'm seeing exactly the same problem. I have about 5000 FLAC and MP3 files, but
no WMA files (so the problem isn't confined to WMA files).

I think this is a fairly serious bug, because on my PC (running Slimserver as a
service on XP SP2), my Squeezebox loses its connnection as soon as Slimserver
hits 100% processor load.  The only way to get things going again is to restart
the Slimserver service, which only works until Slimserver starts to rebuild its
cache again....

Like Stuart says, the 2004-10-03 nightly shows the problem, but 2004-10-02 works
fine.
Comment 5 Stuart Beesley 2004-10-07 15:04:09 UTC
Created attachment 173 [details]
This file on its own causes 100% processor

I've tried lots of other files on their own and this one causes the 100%
problem
Comment 6 KDF 2004-10-07 15:27:15 UTC
good file.  for the record, here's the CPAN debug output:
Working on an ASF_Header_Extension_Object:

        nextObjectGUID: [7C4346A9-EFE0-4BFC-B229-393EDE415C85]
        nextObjectName: [ASF_Language_List_Object]
        nextObjectSize: [39]

        nextObjectGUID: [44231C94-9498-49D1-A141-1D134E457054]
        nextObjectName: [ASF_Metadata_Library_Object]
        nextObjectSize: [374]

        ASF_Metadata_Library_Object: 0
                name  = WM/MediaClassPrimaryID
                value = D1607DBC-E323-4BE2-86A1-48A42A28441E
                type  = 6
                data_length = 16

        ASF_Metadata_Library_Object: 1
                name  = WM/MediaClassSecondaryID
                value = 00000000-0000-0000-0000-000000000000
                type  = 6
                data_length = 16

        ASF_Metadata_Library_Object: 2
                name  = WM/WMContentID
                value = 0B347A08-605F-41B0-9FB5-0251F617A13D
                type  = 6
                data_length = 16

        ASF_Metadata_Library_Object: 3
                name  = WM/WMCollectionID
                value = 7A48D370-0018-4463-B394-2EE4A3E91809
                type  = 6
                data_length = 16

        ASF_Metadata_Library_Object: 4
                name  = WM/WMCollectionGroupID
                value = 7A48D370-0018-4463-B394-2EE4A3E91809
                type  = 6
                data_length = 16

        nextObjectGUID: [00000000-0000-0000-0000-000000000000]
        nextObjectName: [ASF_Unknown_Object]
        nextObjectSize: [0]

        nextObjectGUID: [00000000-0000-0000-0000-000000000000]
        nextObjectName: [ASF_Unknown_Object]
        nextObjectSize: [0]

        nextObjectGUID: [00000000-0000-0000-0000-000000000000]
        nextObjectName: [ASF_Unknown_Object]
        nextObjectSize: [0]

        nextObjectGUID: [00000000-0000-0000-0000-000000000000]
        nextObjectName: [ASF_Unknown_Object]
        nextObjectSize: [0]

        nextObjectGUID: [00000000-0000-0000-0000-000000000000]
        nextObjectName: [ASF_Unknown_Object]
        nextObjectSize: [0]
...loops here for good.

I'll fwd to Dan and see if he can fix it.
Comment 7 KDF 2004-10-07 15:52:14 UTC
adding this at line 508 of CPAN/Audio/WMA.pm seems to alleviate the problem:

last if $nextObjectSize == 0;
Comment 8 KDF 2004-10-07 16:24:16 UTC
Dan checked in a fix for WMA files.  Try tomorrow's nightly and we can close
this off if all is well again.

if it still presists, please try to find the exact file again.  
Comment 9 Stuart Beesley 2004-10-08 11:25:17 UTC
Great!! Scans now back to 30% and 5 mins

Thanks!
Comment 10 Chris Owens 2006-06-16 14:40:37 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.
Comment 11 Chris Owens 2008-12-18 11:51:59 UTC
Routine bug db maintenance; removing old versions which cause confusion.  I apologize for the inconvenience.