Page 1 of 2
Can't fetch NZB from my indexer on specific SABnzbd
Posted: December 24th, 2014, 12:49 pm
by josh4trunks
Hi everyone, I'm trying to grab an NZB from my nZEDb (newznab compatible) install. I'm trying to figure out how I can further debug this and get to the bottom of this issue. Thanks!
PROBLEM
The link i want to work is..
https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
The log shows..
Code: Select all
Grabbing URL https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
2014-12-24 09:34:16,288::DEBUG::[urlgrabber:372] No response from indexer, retry after 60 sec
2014-12-24 09:34:16,289::INFO::[urlgrabber:199] Retry URL https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
OS = FreeBSD v9.3
SABnzbd = v0.7.20
Python = v2.7.8
TESTS
This link works, I uploaded the same nzb to my server
https://www.jruehlig.com/nzedb/test.nzb
This link doesn't work, I temporarily disabled https on my server
http://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
Grabbing an NZB from OZnzb's API, which is also https and newznab based, works.
On a different SABnzbd instance the original link works
OS = FreeBSD v8.3
SABnzbd = v0.7.20
Python = v2.7.2
My tests lead me to believe it is something with the newer Python and my nZEDb api?
Re: Can't fetch NZB from my indexer on specific SABnzbd inst
Posted: December 24th, 2014, 2:05 pm
by shypike
We normally test on Python 2.7, I'm not sure about the last digit.
Do you have a possibility to test on another Python version?
Re: Can't fetch NZB from my indexer on specific SABnzbd inst
Posted: December 24th, 2014, 11:22 pm
by josh4trunks
shypike wrote:We normally test on Python 2.7, I'm not sure about the last digit.
Do you have a possibility to test on another Python version?
I see FreeBSD just added 2.7.9. I'll update and see if that makes a difference.
EDIT
Still doesn't work.
Re: Can't fetch NZB from my indexer on specific SABnzbd inst
Posted: December 25th, 2014, 4:26 am
by sander
When I fill out the URL, I get the same error on my system.
So I added a debug line (logging.debug("SJ: I got in the except due to %s", sys.exc_info()[0])) to my sabnzbd/urlgrabber.py, and the URL results in:
Code: Select all
2014-12-25 10:07:49,039::INFO::[urlgrabber:78] URLGrabber starting up
2014-12-25 10:07:54,044::INFO::[urlgrabber:117] Grabbing URL https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
2014-12-25 10:07:59,012::DEBUG::[urlgrabber:136] SJ: I got in the except due to <class 'urllib.ContentTooShortError'>
2014-12-25 10:07:59,012::DEBUG::[urlgrabber:376] No response from indexer, retry after 60 sec
So: urllib.ContentTooShortError
exception urllib.error.ContentTooShortError(msg, content)
This exception is raised when the urlretrieve() function detects that the amount of the downloaded data is less than the expected amount (given by the Content-Length header). The content attribute stores the downloaded (and supposedly truncated) data.
So a problem with the Content-Lenth.
Re: Can't fetch NZB from my indexer on specific SABnzbd inst
Posted: December 25th, 2014, 4:59 am
by sander
Ah, this is interesting: I can reproduce an error in a few lines of code:
Code: Select all
$ python
Python 2.7.6 (default, Mar 22 2014, 22:59:56)
[GCC 4.8.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import urllib
>>> opener = urllib.FancyURLopener({})
>>> url = 'https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704'
>>> fn, header = opener.retrieve(url)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib/python2.7/urllib.py", line 264, in retrieve
size = int(headers["Content-Length"])
ValueError: invalid literal for int() with base 10: '2795588, 2795588'
>>>
So ... python / urllib can't convert it with int() because it's an array with two int values= (where only one is allowed) which leads to the Content-Length error!
curl output:
Code: Select all
< HTTP/1.1 200 OK
* Server nginx/1.6.2 is not blacklisted
< Server: nginx/1.6.2
< Content-Type: application/x-nzb
< Content-Length: 2795588
< X-Powered-By: PHP/5.6.3
< Set-Cookie: PHPSESSID=44v3sk5mqps0ltremgl406ae87; path=/; secure; HttpOnly
< Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
< Pragma: no-cache
< Content-Disposition: attachment; filename=Slow.Burn.2005.1080p.BluRay.DTS-HD.x264-BARC0DE.nzb
< Expires: Fri, 25 Dec 2015 02:00:06 -0800
< X-DNZB-Category: Movies > HD
< X-DNZB-Details: https://www.jruehlig.com/nzedb/details/fe1a1474a5026b74b5d63b78634f6b055add149a
< X-DNZB-MoreInfo: http://www.imdb.com/title/tt0376196
< X-DNZB-Name: Slow.Burn.2005.1080p.BluRay.DTS-HD.x264-BARC0DE
< X-DNZB-NFO: https://www.jruehlig.com/nzedb/nfo/fe1a1474a5026b74b5d63b78634f6b055add149a
< X-DNZB-RCode: 200
< X-DNZB-RText: OK, NZB content follows.
< Accept-Ranges: bytes
< Date: Thu, 25 Dec 2014 10:00:06 GMT
< X-Varnish: 1713665329
< Age: 0
< Via: 1.1 varnish
< X-Cache: MISS
< Content-Length: 2795588
<
So: twice "Content-Length: 2795588" in the HTTP ... that could be the cause. So:
server side problem. Can you solve that? Do you use a web server behind a proxy server?
Related:
https://bugzilla.redhat.com/show_bug.cgi?id=692525
The part in urllib.py that goes wrong when there is more than 1 Content-Length in headers:
Code: Select all
if "content-length" in headers:
size = int(headers["Content-Length"])
... changing code in a system library on your own system is not something I see as a solution. ;-)
Re: Can't fetch NZB from my indexer on specific SABnzbd inst
Posted: December 25th, 2014, 5:57 am
by sander
Short summary:
Good (with test.nzb just one Content-Length:):
Code: Select all
$ curl -vs -o mynzb.nzb https://www.jruehlig.com/nzedb/test.nzb 2>&1 | grep Content-Length
< Content-Length: 2795588
Bad (resulting in two times "Content-Length:" in the HTTP header):
Code: Select all
$ curl -vs -o mynzb2.nzb 'https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704' 2>&1 | grep Content-Length
< Content-Length: 2795588
< Content-Length: 2795588
So ... solve it on your server side?
Re: Can't fetch NZB from my indexer on specific SABnzbd inst
Posted: December 25th, 2014, 7:29 am
by josh4trunks
Thanks for the debugging! I do have a somewhat complex setup for my hosting.
HAProxy (SSL)
Varnish (Static Caching + GZIPing)
NGINX (Static Webserver)
PHP-FPM
I already tried cutting out the SSL step and that didn't help, so I'll try further isolating the problem. What's strange is how it works on my older setup.
Thanks!
Re: Can't fetch NZB from my indexer on specific SABnzbd inst
Posted: December 25th, 2014, 7:43 am
by sander
Ouch. Well if you're able to construct such a setup, I'm sure you're able solve this tiny Content-Length problem ... ;-)
Re: Can't fetch NZB from my indexer on specific SABnzbd inst
Posted: December 25th, 2014, 12:52 pm
by josh4trunks
sander wrote:Ouch. Well if you're able to construct such a setup, I'm sure you're able solve this tiny Content-Length problem ... ;-)
Looks like Varnish 3.0.4 had a bug which was fixed in 3.0.5 about duplicate Content-Length headers. pfSense's package is only 3.0.4, so time to complain to the package maintainer =]
Thanks, my problem had been solved, move along, and have a Merry Christmas.
Re: Can't fetch NZB from my indexer on specific SABnzbd
Posted: January 3rd, 2015, 9:06 pm
by josh4trunks
@sander
I fixed the Content-Length issue. Are you able to download the nzb's now? I still can't. I'll get some logs posted soon.
Re: Can't fetch NZB from my indexer on specific SABnzbd
Posted: January 4th, 2015, 2:00 am
by sander
It does not work on the stock version of SABnzbd:
Code: Select all
2015-01-04 07:46:42,344::INFO::[urlgrabber:116] Grabbing URL https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
2015-01-04 07:46:46,366::DEBUG::[urlgrabber:372] No response from indexer, retry after 60 sec
2015-01-04 07:46:46,366::INFO::[urlgrabber:199] Retry URL https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
In the 0.7.x git version, it doesn't work, with explanation urllib.ContentTooShortError:
Code: Select all
$ git branch
* 0.7.x
develop
master
2015-01-04 07:56:50,572::INFO::[urlgrabber:117] Grabbing URL https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
2015-01-04 07:56:54,530::DEBUG::[urlgrabber:136] Exception <class 'urllib.ContentTooShortError'> trying to get the url https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
2015-01-04 07:56:54,531::DEBUG::[urlgrabber:375] No response from indexer, retry after 60 sec
2015-01-04 07:56:54,531::INFO::[urlgrabber:202] Retry URL https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
It does work on the 'develop' git version of SABnzbd:
Code: Select all
$ git branch
0.7.x
* develop
master
2015-01-04 07:41:46,990::INFO::[__init__:466] Fetching https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
2015-01-04 07:41:46,991::INFO::[nzbqueue:236] Saving queue
2015-01-04 07:41:46,992::DEBUG::[__init__:809] Saving data for SABnzbd_nzo_u6rO4Q in /home/sander/.sabnzbd/admin/future
2015-01-04 07:41:46,993::INFO::[__init__:884] Saving data for queue10.sab in /home/sander/.sabnzbd/admin/queue10.sab
2015-01-04 07:41:46,994::INFO::[urlgrabber:96] Grabbing URL https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704
2015-01-04 07:41:52,087::INFO::[misc:812] Creating directories: /home/sander/Downloads/incomplete/Slow.VVV.2005.1080p.BluRay.DTS-HD.x264-BARC0DE
2015-01-04 07:41:52,090::DEBUG::[__init__:809] Saving data for SABnzbd_nzf_QK_Z3i in /home/sander/Downloads/incomplete/Slow.VVV.2005.1080p.BluRay.DTS-HD.x264-BARC0DE/__ADMIN__
2015-01-04 07:41:52,090::INFO::[nzbstuff:425] File Slow.VVV.2005.1080p.BluRay.DTS-HD.x264-BARC0DE - [00/63] - "BARC0DE.nzb" yEnc (1/8) added to queue
The small python test snippet does work:
Code: Select all
sander@flappie:~$ python
Python 2.7.6 (default, Mar 22 2014, 22:59:56)
[GCC 4.8.2] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import urllib
>>> opener = urllib.FancyURLopener({})
>>> url = 'https://www.jruehlig.com/nzedb/getnzb/fe1a1474a5026b74b5d63b78634f6b055add149a&i=3&r=de390776ff47336d50e5585876aa1704'
>>> fn, header = opener.retrieve(url)
>>> print header
Server: nginx/1.6.2
Content-Type: application/x-nzb
X-Powered-By: PHP/5.6.3
Set-Cookie: PHPSESSID=ugenhd6lcj43o21b18srlbbl61; path=/; secure; HttpOnly
Cache-Control: no-store, no-cache, must-revalidate, post-check=0, pre-check=0
Pragma: no-cache
Content-Disposition: attachment; filename=Slow.Burn.2005.1080p.BluRay.DTS-HD.x264-BARC0DE.nzb
Expires: Sun, 03 Jan 2016 22:53:24 -0800
X-DNZB-Category: Movies > HD
X-DNZB-Details: https://www.jruehlig.com/nzedb/details/fe1a1474a5026b74b5d63b78634f6b055add149a
X-DNZB-MoreInfo: http://www.imdb.com/title/tt0376196
X-DNZB-Name: Slow.Burn.2005.1080p.BluRay.DTS-HD.x264-BARC0DE
X-DNZB-NFO: https://www.jruehlig.com/nzedb/nfo/fe1a1474a5026b74b5d63b78634f6b055add149a
X-DNZB-RCode: 200
X-DNZB-RText: OK, NZB content follows.
Accept-Ranges: bytes
Date: Sun, 04 Jan 2015 06:53:24 GMT
X-Varnish: 780103128
Age: 0
Via: 1.1 varnish
Connection: close
X-Cache: MISS
Content-Length: 2795588
>>> print fn
/tmp/tmpFkHX0r
>>>
So .. quite a mystery to me. Hopefully not something intermittent ...
Re: Can't fetch NZB from my indexer on specific SABnzbd
Posted: January 4th, 2015, 2:54 am
by josh4trunks
@sander
Thanks for the testing, that log from 0.7.X points me in a direction to test.
I tested querying my webserver directly NGINX > PHP from the install that is having an issue and that "fixes" it so somewhere is the Varnish layer.
I have Varnish GZIP plaintext so I'll double check if this file is GZIPed and if Content-Length is being represented incorrectly.
I know clients (browsers / urllib) are supposed to say what what compression they support (GZIP/DEFLATE) and servers can then choose to do transparent compression, for hopefully compressible files. I'll try to find out what length they are supposed to give (compressed || uncompressed), if my particular setup is doing this wrong or if sabnzbd 0.7.X has a bug for this.
Re: Can't fetch NZB from my indexer on specific SABnzbd
Posted: January 4th, 2015, 10:31 pm
by josh4trunks
@sander
I think I found the issue after reading this
https://sabnzbd.org/viewtopic.php?f=3&t=12992&p=78409
Sabnzbd actually passes the header saying it do gzip encoding.
Varnish then see's this and gives..
Code: Select all
Content-Encoding: gzip
Transfer-Encoding: chunked
Since sabnzbd uses urllib, which is HTTP/1.0, this issue occurs
"if the sender does not include a Content-Length header, the recipient cannot tell if the message has been truncated due to transmission problems."
from..
http://www8.org/w8-papers/5c-protocols/key/key.html
I think my Varnish is setup to assume everyone can handle HTTP/1.1 headers, so in a way it is my problem.
####
So, here's my question =]
* When is sabnzbd switching to urllib2?
Thanks!
Re: Can't fetch NZB from my indexer on specific SABnzbd
Posted: January 5th, 2015, 9:04 am
by shypike
josh4trunks wrote:
* When is sabnzbd switching to urllib2?
It has in 0.8.0 (not out yet)
There won't be a new 0.7.x (at least not for issues like this).
Re: Can't fetch NZB from my indexer on specific SABnzbd
Posted: January 5th, 2015, 1:46 pm
by josh4trunks
shypike wrote:josh4trunks wrote:
* When is sabnzbd switching to urllib2?
It has in 0.8.0 (not out yet)
There won't be a new 0.7.x (at least not for issues like this).
so I take it sabnzbd 0.8.0 will be out soon =]