RAM consumption when ip filter is loaded

[morfikov]

I'm using debian testing distro and I have an issue concerning ip filter list. I tested this using standard qbittorrent client as well as qbittorrent-nox, versions form sid:

Code: Select all

# apt-cache policy qbittorrent
qbittorrent:
  Installed: 3.1.3-1
  Candidate: 3.1.3-1
  Version table:
 *** 3.1.3-1 0
        500 http://ftp.pl.debian.org/debian/ sid/main amd64 Packages
        100 /var/lib/dpkg/status
     3.1.0-1 0
        900 http://ftp.pl.debian.org/debian/ testing/main amd64 Packages
# apt-cache policy qbittorrent-nox
qbittorrent-nox:
  Installed: 3.1.3-1
  Candidate: 3.1.3-1
  Version table:
 *** 3.1.3-1 0
        500 http://ftp.pl.debian.org/debian/ sid/main amd64 Packages
        100 /var/lib/dpkg/status
     3.1.0-1 0
        900 http://ftp.pl.debian.org/debian/ testing/main amd64 Packages

The same settings for both. RAM usage is about 17-20 MIB for nox and 45-50 MiB for standard client. But here, it's a little bit weird because with time the RAM utilization grows, after several minutes it grew almost to 80M I'm not sure if this is end. But we can assume that after the clients start we have 20/50MiB depending on the client. Now, what happens when I add an IP filer list. Shortly after I start the qbittorrent client, it has about 60-70M, but 10 secs after that it grows to 240M. That was standard client, now the nox. After start it's about 30M, 10 secs after that it has 190-200M

This is the log from nox:

Code: Select all

02/12/2013 20:04:07 - Successfully parsed the provided IP filter: 234145 rules were applied.

I tried to load a smaller filter:

Code: Select all

600K,   10k lines		qb ~60M,	gb-nox 22M
2,6M,   50k lines		qb~80M		qb-nox 32M
5,5M,  100k lines		qb ~100M,	qb-noc 80M

The original ip filter file has 12MiB . does anyone know where this 180M+ come from?

[sledgehammer_999]

On the gui try playing around with cache setting. Tools->Options...->Advanced

[Peter]

Not an answer (sorry we don't have spoiler tag, yet):

Just don't use them.
They harm the swarm, and it will harm you the most.

What does it protect people from?
From some random IP ranges that are known for everyone?
If it's a public software, who would even try to log peers from there?

I know what you say now. But how could I protect myself?
Buy a seedbox if you want speed. If you want to protect your ass only and don't care about speed, just stick to BTGuard or VPNs.
But ipfilters... nah.

Seriously I never heard a proper reason why should anyone use them.
It reminds me of the firewalls built into AV products which let anything connect anyway, so there is like no reason to use them.

[morfikov]

So why people use it, and why is there an option in the settings? Personally I don't care, but it's a little bit strange behavior, don't you think?

I tried to set the two options: disk write cache size (32>8>4) and disk cache expiry interval (60>10) , but the second option doesn't change, no matter what I set there, it always resets to 60s. Anyway, sometimes qb client consumes 130M, and sometimes it backs to 240 -- I haven't changed any option so far, except the two above, or actually the one -- and I have no idea what's going on.

[Peter]

Indeed it's strange, and maybe it's a bug (rather, some wrong implementation somewhere)....
But. Why people use it? Because someone implemented this at first. Some people actually spend time to refresh these lists, so they kill p2p even more.
And at the end of the day, they think "oh wow I helped people again." No, they did not.

Why clients implement it? Because people would whine all day. They want their dose of placebo. If they can't get it, the client is lacking.
And you can't prove your point either, because the average Joe won't even know what the heck filters/Peerblock are doing.

He is just like... "dayum, I won't go to jail..." and he installs it. "such a hacker I am, no wonder I can fix mum's computer every time!".

tl;dr: filters and peerblock are placebo, and they only harm p2p communities/yourself.