[linux] sabnzbd 0.5.6 severe memory leak, using 2GB!??! [SOLVED]

[greenfish]

Hi! I've been using sabnzbd for years, so i'm a huge fan, and thanks for the best server side client out there

Right now I updated my server, and of course sabnzbd was updated as well from 0.4.x.x something to 0.5.6.

I loaded a couple of nzbs from binsearch (roughly around 40GB in data) and to my suprise 20 minutes later all of my downloads had failed due to "memory allocation".

Going into my server a quick "free -m and top" showed me a huge increase in memory allocation. sabnzbd is trying to use up all of my ram and when it fails all of my downloads dies of course.

I only have 2GB of ram but that's never been an issue before. To my little knowledge I belive sabnzbd used like less than 400mb on my system on older versions. Also my system rarely if ever had to resort to using swap at least not 40% of my swap memory.

A quick scooping around on this forum it seems like i'm the only one so far with this problem.

I'll be back with a debug log when my server is rebooted after a new kernel upgrade.

Version: (0.5.6)
OS: (archlinux)
Install-type: (linux source)
Skin (if applicable): (Default)
Firewall Software: (none)
Are you using IPV6? (no)
Is the issue reproducible? (yes - everytime I start the daemon)


My System:

Amd Phenom X2 550+
2GB RAM
archlinux kernel 2.6.36
python2 2.7.1-3

sabnzbd log http://paste.pocoo.org/show/312846/
sabnzbd python log http://paste.pocoo.org/show/312844/
sabnzbd error log empty
Showing "top" screenshot:

EDIT There's something very wrong, i'm getting failed on all of my downloads "Download failed - Out of your server's retention?" <---- i'm on giganews, they have over 800 days in retention the file is less than 16 days old.

[rascalli]

What I do not understand is , why do you have so many instances of sabnzbd+ running ?
There should only be 1 running .. maybe that is causing the problem ?

[freakyzoidberg]

I believe those are threads, have about 20 as well on a debian.

greenfish, i am not sure if it s related but check the cache_limit option in the ini file

[shypike]

You log is full if these messages:
Disk error on creating file  /media/DOWNLOAD2/complete/newsgroups/...

SABnzbd cannot write its downloaded files to disk!

You also start SABnzbd with weird parameters:
-f //.sabnzbd.ini -f /home/greenfish/.sabnzbd.ini/sabnzbd.ini -d
Why twice -f? Why the odd names? Should work, but still...

[greenfish]

What I do not understand is , why do you have so many instances of sabnzbd+ running ?
There should only be 1 running .. maybe that is causing the problem ?

Hi! No those are threads (20 connections from giganews).

freakyzoidberg

I believe those are threads, have about 20 as well on a debian.

greenfish, i am not sure if it s related but check the cache_limit option in the ini file

I'll try anyting if it helps, but remember i've never had issues running sabnzbd before. All of my issues started after I went on the 0.5.6 version.

You log is full if these messages:
Disk error on creating file  /media/DOWNLOAD2/complete/newsgroups/...

SABnzbd cannot write its downloaded files to disk!

But why is that? I mean what's causing the errors? I've never had these issues before. The memory leak and those errors started when I upgraded to the 0.5.6 version of sabnzbd. Just to be sure I ran a memtest and my RAM turns out OK, ran smartclt and fsck on my hdd, all came back normal.

You also start SABnzbd with weird parameters:
-f //.sabnzbd.ini -f /home/greenfish/.sabnzbd.ini/sabnzbd.ini -d
Why twice -f? Why the odd names? Should work, but still...

I start the sabnzbd through the /etc/rc.local by executing "sabnzbd -f /home/greenfish/.sabnzbd.ini/sabnzbd.ini -d"
That's the way i've always booted sabnzbd on my server. Is there another way to turn sabnzbd into a daemon?

I'm gonna try executing "sabnzbd" in console and see what happens.

Thanks for your help and input so far guys

20 minutes later .. ram usage is like 520mb that's not too bad compared to before when 99% of all of my RAM was in use!

I'm gonna try putting sabnzbd in daemon mode and see if that's the main culprit on the 0.5.6 on my end.

Here's how I execute the daemon, anone got a better advice?
"I start the sabnzbd through the /etc/rc.local by executing "sabnzbd -f /home/greenfish/.sabnzbd.ini/sabnzbd.ini -d"



UPDATE: I deleted sabnzd 0.5.6, cleaned out all config files, re-mounted all my mountpoints, re-installed sabnzbd 0.5.6 again but this time I used the daemon script provided on this site. 30 minutes later my RAM usage is only 500mb during downloading and around 700mb when i'm extracting. Thanks everyone for your help