SABnzbd Forums

I have exactly the same problem. I have a QNAP with 0.7.19 running for months (and older SABnzbd versions for years) without any problems. I upgraded to 0.7.20 yesterday, and most downloads stop completing. They will just sit there in the queue with a few more MB to be downloaded. I have several normal usenet servers and 1 block server set as a backup server. The problem seems to be if certain articles are not found on a normal server, the backup server is not tried. Only when I disable all normal server and set the backup server as a normal server the downloads will continue and complete.

There is nothing in the logs that indicates a problem. I have reverted back to 0.7.19 to see how it goes.

One thing I can think of is that the QPKG updates SABnzbd to the latest version using this command: So it could leave behind files of old SABnzbd versions that are no longer part of the current version. I did not find any such files between 0.7.19 and 0.7.20 however.

There are no new files in 0.7.20 (or 0.7.19), only changes in existing files.
At least this is true for our distribution tar.gz, I don't know about the package.
I am very interested in any differences in behaviour that you may find.
Please keep me posted.

Downgrading to 0.7.19 didn't help much. I found one download stuck the same way as described above. The status is set to "downloading" while it is actually not.This time a simple pause and unpause helped to get it going again. I have 3 free servers, 1 paid server and 1 block server as backup. I noticed it was using one of the free servers, and then went on to the backup server. So that seems to be working ok this time.

I still think the problem is that the backup server is not used, so SABnzbd does not come to the conclusion the article is missing on all servers, which is needed to discard it and move on.

I have disabled the free servers for now to see if that helps.

Make sure that Config->Switches->Fail on yenc crc Errors is off.

Thanks, I think I enabled that switch right after the upgrade to 0.7.20. The description is When article has a CRC error, try to get it from another server. but I guess that last part is broken?

So far I haven't been able to show that it actually makes mistakes,
but I have my suspicions about it.
Disabling it is safe because 99.9% of such errors are caused by the uploading to the original server.
The feature assumes that syncing between servers can go wrong; which is actually not very likely.
To put it simply: if one server has a CRC error in an article, they all have.

After the unpause in my last test I did notice CRC errors in the logs. The final download had 11 malformed articles.

Are the file sizes right now?
Stalling is another problem all together.
However, it's more likely to occur when more servers are involved.
Suppose three servers time-out, with 3 retries and 120 sec timeout, that's 18 minutes.
The trick is to set a decent time-out, like 30 sec (personally I use 60).
Not too short, because then you may miss more articles.
It seems that many servers need more time to conclude that they don't have an article.

It seems the download no longer stall. I am back at 0.7.20 but with only 1 main server (and 1 backup). When this problem started the downloads would really stall, but just in a timeout. There were no connections active to any server and even after several hours the downloads did not continue or finish.

If this is stable for a few days I will add my free servers again.