SABnzbd Forums

I'm trying to download a few things from usenet and I have the following error in my log:

(article redacted)
2012-04-15 09:03:30,705::INFO::[decoder:131] Unknown Error while decoding <Article: article=###########, bytes=661039, partnum=145, art_id=None>

What is causing this? I've tried two different usenet servers and I'd really like to get more information about what's going on with this... (note: a lot of other posts download fine, but a few of them have errors like above)

Three possibilities:
- The article is corrupt (poster screwed up)
- The article uses an unsupported encoding
- The article doesn't obey Usenet formatting rules (ASCII 16 instead of ASCII 19 + ASCII 16 as end-of-line)

The latter is supported by the latest 0.7.0Beta2 (See Announcements board).

BTW: if it's just a single article and the post has par2 files, this should auto-repair.

Thanks for the information.

How does multiple servers work in this circumstance?

For example, SABnzbd will download the article from the first server but receive an error ("Unknown error while decoding"). Will it attempt to grab it from another server in case the copy on server #1 was bad, but server #2 is OK?

Reading the docs, it looks like multiple servers is _only_ useful if an article is completely missing from one of them.

Thanks shypike!

Most of the time articles are corrupt on ALL servers.
That's because things go wrong when the poster uploads.
Syncing between servers is quite reliable.
Despite that, there is an option in Config->Switches that instructs articles
with bad checksums to be retried at other servers.
Although this does not apply for "unknown encoding".

OK - Thanks again so much for the help.

I guess you could file this as feature request, but could you also have "unknown encoding" be retried at other servers too? (you definitely know more than me, but I would an article could be corrupt and your decoder might return "unknown encoding", but retrying at a different server might return a complete and valid article)

Thanks again!

Yes, the retries could be more universal, but like I said most of the time all servers
deliver the same article anyway.
BTW: a CRC error is only reported when the article is already recognized as a valid yEnc record.