Page 1 of 1
[0.5.4-Linux] Hanging download blocking queue
Posted: October 2nd, 2010, 3:00 am
by Sypher
Something went wrong with an auto downloaded NZB trough RSS.
Even though the age of the files were less then a few hours, it said 35000d.
This faulty download, blocked the whole queue until I deleted it.
Is it possible that such blocked/hanging downloads are somehow skipped or moved to the bottom of the queue when it happens?
Re: [0.5.4-Linux] Hanging download blocking queue
Posted: October 2nd, 2010, 6:19 am
by shypike
If you set "Only get articles for top of the queue" on Config->Switches
to "off", the queue won't block anymore.
Can you email this particular NZB to bugs at sabnzbd.org ?
Re: [0.5.4-Linux] Hanging download blocking queue
Posted: October 2nd, 2010, 10:43 am
by Sypher
Hmm, that option was indeed enabled.
Regarding the NZB, apparently something went wrong at nzbs.org. The filename in backup is:
"index.php!action=getnzb&nzbid=429587&i=XXXXXXX&h=XXXXXXXXXXXXXXXXXX.nzb".
This file is not even a valid NZB, but a HTML page (with the first line empty).
Re: [0.5.4-Linux] Hanging download blocking queue
Posted: October 2nd, 2010, 11:00 am
by shypike
When you lookup the report number on the nzbs.org website, this is what you get:
NZB no longer exists (too old)
wrapped in HTML code, so it displays properly.
Since HTML resembles XML (the format of NZB), there is a small chance
that SABnzbd will treat it as an NZB.
This is indeed what happens in this case.
The trouble that many websites do have programmatic URLs that allow programs
like SABnzbd to retrieve files.
However as soon as something goes wrong they send HTML crap instead of
the official http error codes (you know: like the famous 404 for a missing page).
Re: [0.5.4-Linux] Hanging download blocking queue
Posted: October 2nd, 2010, 1:36 pm
by Sypher
I fully understand what you mean.
Its pretty annoying that sites aren't producing consistent replies.
There are some things in the NZB/XML files which are not exist in HTML, so maybe it is possible for SABnzbd to check for that to make sure these things don't cause problems?
Re: [0.5.4-Linux] Hanging download blocking queue
Posted: October 2nd, 2010, 2:46 pm
by shypike
True, garbage HTML could be treated as just a failed download.
However, some sites do this also when they have temporary problems.
When processing of the received data fails, we retry.
In this case the output of the site is accepted as a degenerate NZB

I'll see if we can improve this a bit in the future.
The next major release will not hang on empty jobs anymore, solving this particular situation.