SABnzbd Forums

HOLY LORD!

I had pages and pages of results for port 9090. I use two different Sabnzbd plugins for Firefox and Chrome. Apparently they were opening ports galore. My next step was going to disable HTTPS and go back to HTTP, that's what I ended up doing and ill keep an eye on the open ports and see if they climb out of control again.

I assume this is an issue with the browser plug-in and not Sabnzbd?

If so, I can report these issues to the plug-in authors.

Maxtastic wrote: I assume this is an issue with the browser plug-in and not Sabnzbd?

It's likely the plugins, but SABnzbd is very vulnerable for DOS attacks, or something that looks like it.
As SABnzbd is not a public web server, we haven't hardened it against overloading.

shypike wrote:

Maxtastic wrote: I assume this is an issue with the browser plug-in and not Sabnzbd?

It's likely the plugins, but SABnzbd is very vulnerable for DOS attacks, or something that looks like it.
As SABnzbd is not a public web server, we haven't hardened it against overloading.

that's what I figured.

I went to the plugin's website to see about filing a bug with the authors, and someone already did a few days ago.

http://code.google.com/p/sabconnectplus ... 20Priority

Maxtastic wrote:HOLY LORD!

I had pages and pages of results for port 9090. I use two different Sabnzbd plugins for Firefox and Chrome. Apparently they were opening ports galore. My next step was going to disable HTTPS and go back to HTTP, that's what I ended up doing and ill keep an eye on the open ports and see if they climb out of control again.

I assume this is an issue with the browser plug-in and not Sabnzbd?

If so, I can report these issues to the plug-in authors.

Which plugins do you use? SABconnect++ and ... ?

I had a look, and it looks like SABconnect++ is written in Javascript (http://code.google.com/searchframe#n3eM ... c&cd=2&sq=) , so quite high-level. It surprises me that a high level language could cause the problem. However, the bug report you mention seems to say it is caused by sabconnect++

Anyway:: what is the other plugin?

And could activate sabconnect++ with HTTPS, and see if you can reproduce the problem. If not, can you only put the other plugin on HTTPS? Hopefully we can pinpoint the plugin.

And: what was the state of all the HTTPS session according to netstat? CLOSE_WAIT, or something else?

Here is an overview of the tcp state diagram: [img]
http://upload.wikimedia.org/wikipedia/e ... ed_new.svg
[/img]

And I wonder if this is Windows problem, or if it can happen on Linux or OSX too...
EDIT: the bug report on http://code.google.com/p/sabconnectplus ... 20Priority is on Debian Linux, so that answers my question. I'm running sabconnect++ over HTTPS against SAB 070 on my Ubuntu Linux, but so far I've not been able to reproduce the problem.

I found an interesting quote

It has been previously established that a socket in a
long-time-lingering CLOSE-WAIT status, is due to one or the other side
of a TCP connection not properly closing its side of the connection when
it is done with it.

... so the cause could be on SABconnect++'s side, but also cherrypy's side ...

If the problem were easy to reproduce, we could use wireshark to see which applications says what.

from one of the other suggestions in the thread, I stopped using Firefox and was only using Chrome. One fun thing with chrome now if you "sign-in' with your Google account info, it will load all of the addons that you have installed elsewhere. I had my home and work box both opening ports and was compounding the issue.

for Firefox I have been using nzbdStatus 1.0.15.1, but for the past week I kept Firefox closed and only had Chrome open, so it was definitively SABconnect++ that was causing the issue.

all of the status were TIME_WAIT, and I swear there were thousands of them. I can pretty easily duplicate the issue, but it takes a few days for the webserver to become unresponsive. I could probably run Wireshark solely on port 9090 and see what is gathered if that would help.

I have 6hours of uptime since my last restart and you can see the current number of sockets open: http://i.imgur.com/IgC7K.png

the interesting part was the only thing I changed when this started to happen was a fresh build if Win7 64bit(moving from XP).

I wonder why these issues do not surface on Win XP. With so many people running on other builds and OSs, I bet its gotta be a Win7 issue.

so for comparison, I disabled SABconnect++ in Chrome and went back to only using NZBstatus for Firefox.

here is my netstat after 2 hours

http://i.imgur.com/kIG7L.png

a bit better. Looks like its the Chrome plugin that's causing the issues, and not something on the Sabnzbd side.