Windows Explorer, Media Player, and big libraries

Thomas Hawk ranted about Media Center the other day. Charlie Owen and Matt Goyer of Microsoft’s Media Center team responded (Matt on his own blog and in comments on Thomas’s blog), and the upshot is that Thomas’s complaints are being taken very seriously.

I’ll have more to say about the MCE part of this post over on Ed Bott’s Media Central, although probably not till next week. But I want to address one of Thomas’s specific complaints here, because it’s more related to Windows in general.

Some background: Thomas has a very, very large digital music library. Last December, when he and I first exchanged details of this problem, Thomas’s library contained 141,000 files. I’m sure it’s larger now.

Thomas says he encounters disk errors when he tries to copy or back up those files:

Windows Explorer sucks. With a large digital library I simply cannot effectively copy files or back files up without having disc errors. Large batch copy jobs are super difficult as one little error aborts the whole job.

Let’s break this down. As Charlie Owen noted in his response, and I can attest, this is not normal behavior. I have 19 hard disks, internal and external, distributed among seven computers in my office. Collectively they represent well over 3 terabytes of storage. I move large numbers of files between computers constantly. I routinely copy the 16,000 files in my music library over network connections between external hard disks, and I don’t get disk errors. Now, if I try to copy a group of files, one of which is in use and locked by a running process, then Windows Explorer will stop. That is a weakness in Windows Explorer that is (1) being fixed in Windows Vista and (2) easily avoided by using third-party file-management tools. (It’s also what I was referring to when I said Thomas had a “legitimate complaint.”) But aside from that known issue, I’ve never encountered the problems Thomas describes. Nor should any properly configured Windows system, Media Center or otherwise.

So why is this happening to Thomas? I’ve read his complaints on this issue and we’ve exchanged some e-mail messages on this topic in the past. Thomas has told me that MP3 files are being randomly corrupted at frequent intervals. This is not normal behavior. It is not caused by Windows or Windows Media Player. There is no reason why any Windows user should get even a single corrupted file. If this happens, it indicates either a hardware problem (such as a buggy USB controller), a bad configuration (like cached writes being lost during copies), or data being damaged by software.

If I were Thomas, I would do the following things:

First, I would run a thorough diagnostic tool (like Ontrack’s Data Advisor) on all of the hard drives that were giving me problems.

Second, I would convince one of my buddies at Microsoft to put me in touch with an engineer who could verify that all drivers in my storage subsystem were working properly. If necessary, I would have that engineer hook up a remote debugger and then start the copy process until it fails, so that the exact error could be captured.

Third, I would find an MP3 diagnostic utility and check all of my MP3 files to see if any of them have damaged tags. The MP3 file format is flaky, especially in files that use the older ID3v1 format. If the file format is damaged, it could be causing problems during copies. I suspect that the WMP option to save star ratings to music files might be at least partially to blame for this problem. (In fact, I would recommend that Thomas use a batch MP3-editing program to translate all ID3v1 tags to ID3v2 and then rewrite all tags. This would be time-consuming but would have long-term benefits.)

Fourth, I would temporarily disable all features in WMP 10 that update or change music file metadata, especially those that affect star ratings. (This would not have any impact on currently stored ratings.) I would also disable folder monitoring temporarily.

Somewhere in that process, I’m sure the real cause of this problem would become apparent. Is this a lot of work? Well, yes, but this is also an absolutely certain way to fix the problem once and for all.

8 thoughts on “Windows Explorer, Media Player, and big libraries

  1. Ed. Great advice. Thank you very much. Not sure why you and Charlie have it so easy but do wish I had your setups. My 141,000 collection is now down to about 126,000 due to lost files due to the difficulty I’ve had with backing up my collection. Unfortunate.

    I will check out Ontrak when I have some time. I’m not sure option number two makes the most sense. If it were an isolated incident on my PC I’d suspect that could be the problem. But I’ve got the same issues on all four of my PCs that I have at the house. Furthermore, my primary back up for my collection is my brother. He and I have built this collection together and serve as each other’s back up. He also gets copy errors on his PC at his home. I’m currently missing all artists I-O in my collection because a drive went down. My brother is having a hell of a time trying to get copies of I-O back for me for my collection.

    It would seem to me that the problem would be with the hard drives or with the files themselves. As we are using multiple brand hard drives (Seagate, LaCie (ugh), Maxtor and Cobra) I would lean towards the data as the culprit.

    Great advice on the mp3 tag utility and I’ll look into this as well. I do rely on the star ratings extensively and on WMP to update my album info where missing. I will look into this. It seems that even when I identify the offending album and skip that file, the next time I back up the collection it’s the same problem with new different albums.

    It’s great to hear that Vista addresses the batch copy problem with Windows Explorer. I’ve got a copy coming in the mail and perhaps this will be my salvation in the end. If on the test PC where I install it I can now do large batch copy jobs and have Windows simply skip the offending files then it will be problem solved (perhaps) and I can use the Vista PC to do my backups.

    It’s easy to be bitter when you lose large chunks of your music library. For whatever reason in general I’ve found more trouble with external drives than internal drives. I’m hoping that once 64bit computing / Vista is stable that I’ll be able to upgrade my primary PC and get a couple terrabytes internally to store all of my music (hopefully possible next year).

    Thanks again for the advice. I’ll look into it when I have time and report back.

  2. Thomas,

    Your point about internal drives versus external would be better stated as direct bus-level interfaces versus indirect connections. By adding the USB connection in there, you’re forcing the data to go through the USB layer first, then be translated to an IDE connection. Removing that step in the process would significantly increase reliability and performance, IMO.

    I’ve done a little more testing and I am just not convinced that USB drives will ever be reliable enough for what you want to do. Too many weak points in the chain. Given the nature of your data, I would strongly recommend that you consider getting an add-on SATA card with at least two external ports and then upgrading your PC with one or two internal SATA drives and two external SATA drives. Because these drives use native bus level connection as opposed to an IDE bridge, you will get better performance and better reliability.

  3. OK here is the situation I have a large collection of mp3 files on my server. I also have MCE running in another room that points to the MP3,s everything great so far. I exhaustingly retagged all my files with mp3tag, Musicmatch and cover art downloader. Once I had them all just the way I wanted them I then went to my MCE computer and added them to WMP10 so the MCE GUI would see them. Well WMP10 stripped about a hundred mp3 files of their tags and put them into a Unknown genre. That is crazy so I turned off every option in WMP10 that has anything to do with tagging files. Fixed all my MP3,s again and re added them to the WMP10 guess what it did it again stripped about a hundred of all tags and put them all into a Unknown genre and unknown album arghh. So I fixed them again this time making them all read only attributes then re-added them again doggone it WMP10 still did it again but the weird thing is inside the GUI of MCE all the genre’s are correct only now in WMP10 does it show them incorrect. So what I want to know is how the heck can I disable WMP10 from messing with my already tagged mp3,s I hate making them read only or using NTFS to make them read only to the MCE machine because then I have to mess with all those permissions if I want to edit any album art or tags. Is there a solution and has anyone else seen this problem? I just want WMP10 to leave all tags alone.

  4. @Colby:
    I have had this problem with WMP9 where it eats my tags.
    (FYI, I came over to this site on a google search looking fora fix to totally lock it out of touching my tags).
    I have in the options everything related to changing my data but it still does it. I whatched it do it hafter it played a track of an untagged album, it tagged it. (Did it wrong too, and I don’t know how it got the info, prolly from the filename, b/c it dosent have a connection to the net.)

    The only way I have fixed this issue is to lock out WMP from even thinking out touching the files by adding a ‘wmplayer’ user with very minimal permissions to my music collection (~=30gig) and using the “Run as…” right click function and running it under it’s own user.

  5. I have the same problem and more. I have 10,000 mp3. when i utilize wmp10 to organize it lists all songs with greatest hits as album title under one artist.

    To explain:

    Many mp3 are greatest hits as i have named them (album titles).

    wmp searches music and mislables approx. half of mp3’s

    albums are listed twice as greatest hits and then greatest hits (311). this happens for EVERY Artist.

    Then to make matters worse it appears that i have lost mp3’s however if i find the ONE ARTIST that WMP10 has decided to pick on its own. so under greatest hits album for every artist there is nothing or it has magically stripped the album info.

    it them dumps all artist no matter what title in 311 greatest hits?????????

    this is crappy software, same as always with microsoft half assed at best.

    now, i have renamed every file with id 2 tags with id tag scanner and id tag it 3. I labeled each song 10,000 with both software programs to ensure proper labeling before using wmp10. still not labeling properly.

    i have wmp to not update songs titles and unchecked star ratings. unplugged itnernet to be sure (tried it with internet too same result)

    anyway you look at it it screws up my id tags so that i have to rename everything. this truly sucks, it takes a day to restore over 100 gigs.

    anyway if you fix the bug and typ. ms software let me know

    thanks
    mark

  6. Dear Ed,

    Some of my music files have been deleted.

    I have tried all search metheds to no avail. Please help.

Comments are closed.