What to do with CRC errors?

[guitarboy]

I'm a little unclear on what to do when I get a CRC error.  I read in a another post that these are quite rare, but I seem to get them in about 1/10 of my downloads.

Here's an example of what appeared in the history file:

Download
    Downloaded in 50 minutes 29 seconds at an average of 1.53 MB/s
Repair
    [myfile] Quick Check OK
Unpack
    [myfile] ERROR: CRC failed in "Myfile 720p.mkv"
Category
    movies
Path
    k:\downloads\complete\Movies\Myfile

The files are all recent posts and my server (NewsDemon) seems to have decent availability.

If this is a normal occurrence (i.e. 10% of downloads) should I just look for a different post of the same file, or is some way to force post processing of files that are getting this error?

I will say that I have much fewer problems with SABnzbd than with my previous reader NewsRover.

[ncjok]

Myself also. I personally wonder whether I am following an appropriate workflow in these cases; my current approach is very basic:

1) I try to identify the 'corrupt' RAR either visually when it's filesize stands out as being different or using the HashCheck shell extension in conjuction with an included SFV file.
2) I return to the indexing site and create a new NZB containing that RAR and the 'biggest' PAR2 file.
3a) If on second download the RAR is good I have to manually move the RAR into the existing rar-set directory, replacing the corrupt one and run an unrar program myself.
3b) If step 2 again yielded a corrupted RAR I go off to see if it was reposted to Usenet.

Am I correctly using the PAR2 scheme in step 2? In the case above I will have one RAR and one PAR2 file in C:\downloads\complete\release rarfix\ and everything else (for example, nine RARs, one NFO, one SFV and one SRR) in C:\downloads\complete\release\. Are PAR2 files there to allow repair of individual RARs on their own or am I supposed to make sure all files reside in a common directory first?

[shypike]

par2 files add redundancy.
SABnzbd uses them to verify correctness of the actual download
and to repair damaged files.

More info, see: http://forums.sabnzbd.org/index.php?topic=3409.0

[inpheaux]

Repair
    [myfile] Quick Check OK
Unpack
    [myfile] ERROR: CRC failed in "Myfile 720p.mkv"

If quick check passed but you got a CRC error in unpack then you're likely seeing Case 3 from this thread, where the original rar set is broken and the pars created for the rar set were based on broken data. In this case your only option is to download something else.

I highly doubt that 10% of posts you download are resulting in CRC errors for this specific issue, others may be due to CRC error producing issues. If they're all like that (Par2 ok, Rar fail), then there's a good chance you're instead seeing Shypike's Case 4, where your ram is bad. At which point, check it out with memtest.

[inpheaux]

Thank you for your last comment inpheaux. I was just about to post with a query regarding downloads which pass Quick Check but fail RAR extraction. The example I was going to cite can be seen in this screenshot (redacted) where part 36 is clearly undersized and out of place. It was then noted that a repost for that file had been made the next day.

[guitarboy]

Thanks - great support all around.  I've turned on Post-Process Only Verified Jobs - sounds like that will weed out the broken RAR sets.

[shypike]

Not always.
If a poster pars an already corrupt rar-set, it will still pass.
But that's an issue that cannot be prevented.

[inpheaux]

Thanks - great support all around.  I've turned on Post-Process Only Verified Jobs - sounds like that will weed out the broken RAR sets.

In addition to what shypike said, this won't actually fix your problem. It'll stop reporting them as CRC errors and instead switch to the job failing due to par2 failing. It's still a failure, but it'll fail faster. If you continue having par2 fail either your host sucks or your ram is bad.

[Jan_Papst]

A number of reasons could be the factors of CRC error. Whenever a CRC error will be shown on your monitor screen, it means the data inside your archive file has been corrupted. Generally, the corruption occurs during the transfer of data over the internet if the internet connectivity is not good or suffered from unwanted interference or sometimes due to more downloads at the same of time.
But there is an easy task to fix it; you can repair the corrupt archive by Ctrl+P. But in that if it doesn’t work, you have to download the corrupt files again from the original source. And what method would be better than this if you do have a backup? You can restore all the data from the backup file created for that archive file.
Sometimes, CRC errors appear when the target drive is totally full. So always make sure that your drive in which you are going to extract the files should have enough space.
And when any manual trick is not getting the problem fixed for you, then the ultimate solution which has to be select is software tool. You can use a software application that can easily fix the CRC error. Or you can also check the trial version to be more assured with it.
http://www.sysinfotools.com/recovery/archive-recovery.html