bad performance: bug or missing optimisation?
A message that is not in the archive (sent off-list or not yet received): [email protected]
In message <[email protected]> you write:
Quoted 7 lines
Hello!
While fighting with ICMP flood we came across strange NAT problem.
When box is forwarding ICMP flood from one source address to random destination addresses, it becomes unresponsive and CPU usage goes up.
Quoted 6 lines
No NAT was taking place, just conntrack and nat modules loaded.
Some debugging and profiling showed problem is that entry in ip_nat_core.c:bysource for flooding source address hash is growing very large and
find_appropriate_srcis walking all this list and always failing.
Ah, there are a couple of problems here.
The code is trying to ensure that we are compliant with Dan Kegel's proposed NAT RFC changes:
http://www.alumni.caltech.edu/~dank/peer-nat.html
But there are several problems. The first is that your application is probably not varying the ICMP echo ID, so it's hitting a single hash bucket for everything. The second is that the in_range code seems to have a logic bug:
if ((mr->range[i].flags & IP_NAT_RANGE_PROTO_SPECIFIED) && proto->in_range(&newtuple, IP_NAT_MANIP_SRC, &mr->range[i].min, &mr->range[i].max)) return 1;
This should be:
if (!(mr->range[i].flags & IP_NAT_RANGE_PROTO_SPECIFIED) || proto->in_range(&newtuple, IP_NAT_MANIP_SRC, &mr->range[i].min, &mr->range[i].max)) return 1;
Which means it will match the first entry...
Does that help?
Rusty.
--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
Hello,
Quoted 1 line
Quoted 1 line
While fighting with ICMP flood we came across strange NAT problem.
<snip>
Quoted 1 line
Quoted 1 line
No NAT was taking place, just conntrack and nat modules loaded.
<snip>
Quoted 1 line
Ah, there are a couple of problems here.
Do you remeber what I reported some time ago about a problem with ICMP redirects. When the NAT module was loaded, after some time, the kernel stopped reacting to ICMP redirects sent by one of my routers. When i unloaded the NAT module, everything came back to normal, and the kernel reacted to redirects and used to send the packets to the proper IPs.
Has that been investigated?
Regards,
Maciej
In message <[email protected]> you write:
Quoted 5 lines
Do you remeber what I reported some time ago about a problem with ICMP redirects. When the NAT module was loaded, after some time, the kernel stopped reacting to ICMP redirects sent by one of my routers. When i unloaded the NAT module, everything came back to normal, and the kernel reacted to redirects and used to send the packets to the proper IPs.
Hi Maciej,
Definitely a different problem. ICMP errors are handled
very differently from pings. I assume you were actually doing NAT at the time?
Rusty.
--
Anyone who quotes me in their sig is an idiot. -- Rusty Russell.
Hi,
Quoted 5 lines
Hi Maciej,
Definitely a different problem. ICMP errors are handled
very differently from pings. I assume you were actually doing NAT at the time?
No NAT rules, yet the nat modules were simply loaded because i used a copy of a setup that just had:
# iptables -X -t nat # iptables -F -t nat For flushing all the tables. So iptable_nat was in memory. After unloading that, the redirects were treated again by the conntrack mechanism as valid and accepted by the by the ruleset. After modprobing iptable_nat again, after a few minutes, the -m state rules did not match the redirects, so they were DROPped by the ruleset default policy.
I will re-examine the problem with the latest 2.4 kernel.
Regards,
Maciej