INTERACT FORUM

Please login or register.

Login with username, password and session length
Advanced search  
Pages: [1]   Go Down

Author Topic: Testing library integrity  (Read 2719 times)

hit_ny

  • Citizen of the Universe
  • *****
  • Posts: 3310
  • nothing more to say...
Testing library integrity
« on: March 14, 2009, 03:42:31 pm »

- Auto-Import runs a lot faster (I can't remember exactly but I would guess at least 4x faster). This is great because manually running auto-import is the only method for testing library integrity and I run it at least once a day.

Don't follow  ?
Logged

rjm

  • Regular Member
  • Citizen of the Universe
  • *****
  • Posts: 2699
Re: Testing library integrity
« Reply #1 on: March 14, 2009, 06:53:47 pm »

When I run auto-import (which points to all my media roots) and it completes with the message "Imported 0 new files" this tells me:

1) none of my media files have been unintentionally deleted or moved or renamed
2) none of my media file dates, sizes, and tags have been unintentionally changed
3) no body or no thing has added any media files to my library
4) a good chunk of the MC library data (index, file paths, thumbnails, etc.) has not had any unintentional changes
5) I did not forget to import and tag any media
6) I did not forget to build any thumbnails
7) there are no gross problems with my hard drives
8 there are no gross problems with new versions of MC

It's not perfect but it's pretty good at detecting problems and it's all we have got.

It has caught many problems for me including:

1) hard drive problems
2) nasty 3rd party software (like Windows Media Player) changing my media tags without my permission
3) bugs in new versions of MC like when MC12 randomly detected changes in document media types
4) mistakes I have made
Logged

hit_ny

  • Citizen of the Universe
  • *****
  • Posts: 3310
  • nothing more to say...
Re: Testing library integrity
« Reply #2 on: March 15, 2009, 10:14:17 am »

The approach here is, starting with the HD, check if it exists in the library. If not import & report. Catches stuff thats on the HD but not in the library.

The inverse would be to say, start with the library and then check each entry exists on the HD. A smartlist with  [=IsMissing([Filename])]=1  will do the job.

But what about the case if anything vanishes off the HD but whose entry is still in the library ?

I'm thinking only the second approach will catch it  ?

3) no body or no thing has added any media files to my library
I will assume you mean another person here, a single MC instance is shared with others, which is just a recipe for trouble but then thats just me ;)

6) I did not forget to build any thumbnails
Don't follow how an auto-import also builds thumbnails ?

It's not perfect but it's pretty good at detecting problems and it's all we have got.

It has caught many problems for me including:

1) hard drive problems

Could you tell us more :)
Logged

rjm

  • Regular Member
  • Citizen of the Universe
  • *****
  • Posts: 2699
Re: Testing library integrity
« Reply #3 on: March 15, 2009, 11:01:33 am »

The approach here is, starting with the HD, check if it exists in the library. If not import & report. Catches stuff thats on the HD but not in the library.
correct

Quote
The inverse would be to say, start with the library and then check each entry exists on the HD. A smartlist with  [=IsMissing([Filename])]=1  will do the job.

But what about the case if anything vanishes off the HD but whose entry is still in the library ?

I'm thinking only the second approach will catch it  ?
Thanks, I had not thought of using a smartlist. I have set the auto-import options "Update for external changes" and "Fix broken links" so by running auto-import I achieve both "check if file exists in library" and "check if entry exists on HD" in one step. In case you have not tried it, auto-import gives a nice report at the end for anything that is missing or that has changed.

Quote
I will assume you mean another person here, a single MC instance is shared with others, which is just a recipe for trouble but then thats just me ;)
I have a complicated network with many shared drives. My kids have access to all the media and use a mix of tools for accessing it including JRiver, Windows Media Center, iTunes, and Orb. I have tried to design the network so that they have read-only access and cannot change anything however in case I made a mistake in setting up the network, or in case an application does not respect the read-only status (like MC does when rotating photos) my auto-import scheme will let me know if anything has changed.

Quote
Don't follow how an auto-import also builds thumbnails ?
I have set the auto-import option "Build thumbnails". If you watch the auto-import progress in the Action Window you will see that at the end of scanning all the drives it reports building thumbnails if any need to be built. So if I do not see this message then I know I did not forget to build any thumbnails. If I did forget to build some thumbnails then there is a high probability that I also forgot to tag and rename the files so I go have a look to see if I made a mistake.

Quote
Could you tell us more :)
I have about 4 TB of media spread across 6 drives. All of this is backed up to another set of external and network drives. Because I frequently run auto-import as well as my backup program which scans for changes I have developed a good sense of how reliable hard drives and/or Windows are. My experience has been that about once every 2 months I see one or two random files (out of 193,000) change. Usually the change is a single character in the file name. I have no idea what causes this. Usually a reboot cleans up the problem but sometimes it does not and I have to restore the file from the backup.

The lesson is that you cannot trust that your media is safe just because you are not getting any drive error messages from Windows. You need to check the data every once in a while and MC's auto-import is one great way of doing this.
Logged

leezer3

  • MC Beta Team
  • Citizen of the Universe
  • *****
  • Posts: 1587
Re: Testing library integrity
« Reply #4 on: March 15, 2009, 01:59:41 pm »

I have about 4 TB of media spread across 6 drives. All of this is backed up to another set of external and network drives. Because I frequently run auto-import as well as my backup program which scans for changes I have developed a good sense of how reliable hard drives and/or Windows are. My experience has been that about once every 2 months I see one or two random files (out of 193,000) change. Usually the change is a single character in the file name. I have no idea what causes this. Usually a reboot cleans up the problem but sometimes it does not and I have to restore the file from the backup.

The lesson is that you cannot trust that your media is safe just because you are not getting any drive error messages from Windows. You need to check the data every once in a while and MC's auto-import is one great way of doing this.


Never, ever seen anything like that. I've got well over 4.5TB of stuff (Approx 225,000 files imported into MC last time I bothered to check, this spans 14 drives of various ages & sizes), which has collected over the past 7 years or so.
One of the drives in my server is the 20gb original from a Windows 98 box, and everything is still precisely where I left it.
For that matter, the only drives I've ever had die (2x Samsung 400gb Spinpoints) were sickly from the start, and they never had any important data on them :)
My personal approach on this front is to use any new drive as the general download/ data dump. The drive that it replaces then moves into the main storage pool after having been running constantly for 6 months or so, which gives enough time for any imminent failures to become apparent.

Admittedly, my main PC and the network clients are the only people accessing this data, so there's no real external error factor involved, which is present in your setup.
Even with this in mind, if you're having filenames randomly change then TBQH you have hardware issues somewhere. I'm not going to speculate on what these might be, but something is up.

Cheers

-Leezer-
Logged

hit_ny

  • Citizen of the Universe
  • *****
  • Posts: 3310
  • nothing more to say...
Re: Testing library integrity
« Reply #5 on: March 15, 2009, 02:52:48 pm »

The lesson is that you cannot trust that your media is safe just because you are not getting any drive error messages from Windows. You need to check the data every once in a while and MC's auto-import is one great way of doing this.

Now auto-import is great way to check if tags got externally updated, implying either a reduction or increase in file size as well as date changes.

But what about the case where the file size remains the same ?

If a HD is progressively going bad, now if whole files will go missing, you will be warned. But when select bytes go bad whilst leaving the whole file detectable then the only way to know is checksums.

But they come with a cost, its a lot more involved than just clicking auto-import as well as the caveat that you can't update file tags once checksums are generated.
Logged

rjm

  • Regular Member
  • Citizen of the Universe
  • *****
  • Posts: 2699
Re: Testing library integrity
« Reply #6 on: March 15, 2009, 02:58:23 pm »

Never, ever seen anything like that. I've got well over 4.5TB of stuff (Approx 225,000 files imported into MC last time I bothered to check, this spans 14 drives of various ages & sizes), which has collected over the past 7 years or so.
One of the drives in my server is the 20gb original from a Windows 98 box, and everything is still precisely where I left it.
For that matter, the only drives I've ever had die (2x Samsung 400gb Spinpoints) were sickly from the start, and they never had any important data on them :)
My personal approach on this front is to use any new drive as the general download/ data dump. The drive that it replaces then moves into the main storage pool after having been running constantly for 6 months or so, which gives enough time for any imminent failures to become apparent.

Admittedly, my main PC and the network clients are the only people accessing this data, so there's no real external error factor involved, which is present in your setup.
Even with this in mind, if you're having filenames randomly change then TBQH you have hardware issues somewhere. I'm not going to speculate on what these might be, but something is up.

Cheers

-Leezer-
Yes I agree, probably some infrequent hardware problem - maybe a bit of ram flipping once in a while - I don't know. It happens so infrequently that it is impossible for me to diagnose.

I should have mentioned that a full binary comparison between main and backup drives is also a good idea once in a while. You learn strange things by doing this. For example, Microsoft Excel modifies the xls file every time you open it, even if you do not save the file. In other words, the content of an Excel file changes everytime you open it, even though the last modified date and size do not change. Very bad practice in my opinion.
Logged

rjm

  • Regular Member
  • Citizen of the Universe
  • *****
  • Posts: 2699
Re: Testing library integrity
« Reply #7 on: March 15, 2009, 03:02:18 pm »

Now auto-import is great way to check if tags got externally updated, implying either a reduction or increase in file size as well as date changes.

But what about the case where the file size remains the same ?

If a HD is progressively going bad, now if whole files will go missing, you will be warned. But when select bytes go bad whilst leaving the whole file detectable then the only way to know is checksums.

But they come with a cost, its a lot more involved than just clicking auto-import as well as the caveat that you can't update file tags once checksums are generated.
Yes, I do a binary comparison against the backup drives once in a while.
Logged
Pages: [1]   Go Up